hash
stringlengths 32
32
| doc_id
stringlengths 5
12
| section
stringlengths 4
1.47k
| content
stringlengths 0
6.67M
|
|---|---|---|---|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
4 Overall description of voice over IP in the IMS domain when connected to GERAN
|
GERAN is considering the solution to provide an optimized voice bearer as well as generic bearers to support speech originating from the Iu-ps. The optimization is achieved by reusing the channel coding of CS speech channels in GSM, and by employing header removal to increase the spectrum efficiency. The consideration regarding header removal was made with the understanding that header removal is a non-transparent header adaptation scheme and that therefore optimized voice can’t be used together with synchronized medias.
Optimized voice will be used in conjunction with SIP. Agreed schemes in GERAN to transport SIP are DTM (Dual transfer mode: going over to 2 half rate or full rate slots during the transmission of SIP data) or FACCH, stealing speech frames during the SIP transmission periods. Both schemes are already provided by GSM R99 or earlier.
5 Definition of optimized voice schemes
5.1 Header Removal
Transport and network level headers (e.g. RTP/UDP/IP) are completely removed. Based on information submitted at call set-up and based on information derived from lower layer (link & physical), the receiving entity can regenerate the headers. The primary application of header removal is the optimized speech bearer, and the regenerated header may not always be semantically identical to the original header.
5.2 Header Compression
Transport and network level headers (e.g. RTP/UDP/IP) are compressed in such a way that the decompressed headers are semantically identical to the original uncompressed headers. The IETF ROHC WG is responsible for standardising header compression schemes. Header compression is suited for standard Internet applications that are not designed to work only with GERAN and especially for multimedia applications therefore the scheme will be used with generic real time multimedia bearers.
6 Requirements and working assumptions for support of voice optimisation for the IMS in the GERAN
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
6.1 Requirements
|
1. It shall be possible to use a SIP based optimised voice service with a mobile terminal supporting multi slot class 1 (1 TS in DL, 1 TS in UL).
2. There shall be no performance degradation in coding and modulation compared to traditional circuit switched voice services.
3. GERAN shall not interpret SIP messages.
4. The GERAN solution shall utilise as far as possible already existing protocol means on the Iu interface for UTRAN.
4.1 Although UTRAN has no plans to deploy header removal in Rel 5, a solution shall take into consideration UTRAN developments and UTRAN architectural principles.
5. The change between header compression and header removal shall be possible during handover.
6. Interruptions in speech due to SIP signalling, mid call, shall be kept to a minimum. SIP compression is required.
7. The GERAN solution shall be future proof and shall not exclude the support of multiple codecs.
8.Whether header regeneration is carried out in the MS shall be an implementation issue
9. It shall be possible to identify whether the terminals has requested an optimized voice bearer or a generic radio bearer for carrying voice in the call data records (CDR).
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
6.2 Working assumptions
|
1. It is unclear when/whether mid path transcoders for the IMS will be available between two SIP end users.
2. TSG GERAN is responsible to develop the header removal solution for an Optimized Voice bearer.
3. GERAN informs the MS which codecs are currently supported and the MS is in charge of identifying a single codec, which is supported by the GERAN and the other SIP endpoint (FFS). The mobile requests resources from the network.
4. The GERAN will make the final decision whether or not header removal is possible to apply, or if a generic radio bearer will have to be used.
5. It will not be possible to use header removal for bearers that are part of a multimedia session requiring synchronised media streams.
6. As RTP time stamps and sequence numbers are generated in the BSS, thus there might be an offset in the generated headers across a handover event. Positive or negative slips in sequence numbers may occur in such a situation.
7. In initial implementation it is assumed that the application that generates and receives the flow for which header removal is applied, is integrated in the terminal. Refer to 3.1.1 for the definition of an application that is integrated in the terminal.
8. Header removal cannot be used where end-to-end encryption or integrity protection is used as it does not guarantee bit-exact transfer of traffic.
7 Issues for the support of header removal within GERAN
The purpose with the following subchapters is to capture all issues related to the support of header removal within GERAN. Each subchapter is in turn divided into subchapters describing the characteristics of the problem, possible solutions and the working assumptions that have been agreed.
When a working assumption has been adopted, the solutions that have not been chosen are not removed. The reason for this approach is to avoid that discussions around matters that already have been concluded, shall pop up again at a later stage.
Figure 1. The figure illustrates the process when an Iu-PS voice call is set-up in GERAN. The overall principles are, where nothing else stated, basically the same in all solutions described in chapter 7.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.0.1 Summary of GERAN/MS States for IMS Calls with Header Removal
|
A number of events have been identified. These are:
1. At some point, MS is made aware of current local Codec/Channel Coding Support
Note: This information may no longer be accurate by the time that the GERAN needs to allocate a bearer, due to changes in resource availability
2. MS has PDP Context for SIP traffic
3. MS has engaged in SIP signalling; a Secondary PDP Context Activation has not yet been requested
4. MS sends the Secondary PDP Context Activation Request; SGSN sends an associated RAB Assignment Request; BSC has not selected RB and channel coding scheme yet
5. BSC selects the final RB and channel coding scheme, and initialises the PDCP entities as part of the RB (or extended RB) setup procedure. There are two variants here
5.1 Header Removal is used
Note: This event defines the “latest point” at which IP address/port and Payload Type information must be available at the BSC in order for Header Removal to be initialised, regardless of the technique used to deliver the information.
5.2 Header Compression (or no adaptation) is used.
6. RB Setup is complete, but media traffic has yet to begin
Note: It is assumed that SIP call setup is complete before media traffic transfer begins
7. Media traffic transfer is active; the access link is stable
8. Handover occurs whilst maintaining the current PDCP mode (and the same Codec/channel coding scheme). Again, there are two variants here;
8.1 Header Removal is used
8.2 Header Compression (or no adaptation) is used.
9. Handover occurs whilst maintaining Header Removal, but involving a change in Codec and channel coding scheme (including a change in ACS whilst maintaining AMR as the Codec)
10. Handover occurs, together with a PDCP mode change from Header Removal to Header Compression. There are two variants here:
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
10.1 PDCP Mode change from HR to HC is part of an Inter-RAT change from GERAN to UTRAN
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
10.2 PDCP HR to HC mode change occurs within GERAN
|
11. Handover occurs, together with a PDCP mode change from Header Compression to Header Removal. This has two variants:
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
11.1 There is also an Inter-RAT change from UTRAN to GERAN
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
11.2 PDCP mode change without such an Inter-RAT change"
|
Note: Some of these events (notably those involving complex handover cases) are not covered within the Technical Report.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.0.2 PASNAS Information Structure
|
A group of data items can be used to assist the PDCP processor in selecting the appropriate scheme to be used when performing packet adaptation within the RAN. It can be viewed as a set of “suggestions” from the MS to the RAN. This information is doubly optional:
7 the MS need not send it (relying on default behaviour from the RAN),
8 the RAN need not act on it (either due to the information not being appropriate in the particular configuration, or because the requirements are not applicable to the processing it will perform).
This group of Packet Adaptation Specific Non-Access-Stratum (PASNAS) Information is intended to be valid for a given bearer and applicable over the lifetime of that bearer. Other information may also be needed to specify fully the operation of Packet Adaptation in the RAN; this need not be transferred in the same way, or even at the same time.
Data specifying the traffic type to be carried within a bearer used in a multimedia call is RAN-specific, and is needed only to improve the efficiency of RAN-based packet adaptation. It should not have an impact on the operation of the Core Network, and should not need to be modified or read by the Core Network, even if this is used to relay the information between the MS and the RAN. As a result, this information should be carried inside a Transparent Container whilst being relayed via the Core Network.
Such an (optional) PASNAS Information structure will contain a set of fields, with two initial field types being defined. If the PASNAS Information has not been provided by the MS, then the GERAN will have to assume a conservative approach to header adaptation (i.e. Header Removal will not be possible).
The possible field types that may be contained in such a structure are specified next.
7.0.2.1. Adaptation Type Requirements
It is proposed that a bit set be used to hold flags indicating the mobile application’s special requirements on the adaptation scheme to be used for the associated bearer.
There are currently two situations in which a mobile application may want special treatment of packets being carried through a RAN. These are reflected in the following flags. There may be other flags added in the future, but these two cover the initial needs for GERAN and may be useful for the UTRAN case as well. It is expected that additions to these flags will be restricted, and so the Adaptation Type Requirements field can be of a fixed size (e.g. the bit set will fit into a single octet, for a total field size of two octets, including the tag).
If all these flags are set to false, the GERAN should interpret this as an explicit statement by the mobile application that it is no specific adaptation type requirements. In this case, Header Removal is allowed.
7.0.2.1.1. Synchronisation Indicator
Header Removal is not possible within a GERAN if the speech media flow is part of a multimedia application requiring synchronisation between the different media flows (see Section 7.5.1 of the TR on Optimised Voice). Thus, one reason why Header Removal might mot be allowed by the mobile application is that the associated bearer is to carry such a “synchronised stream”. The RAN will be unaware of this fact, and so, if the mobile application requires special treatment for this flow, it will have to indicate this, using a “Synchronisation Indicator” flag.
More generally, it should still be possible to use Header Adaptation where a bearer is so indicated; however, the RAN should not use adaptation mechanisms that will make it difficult for such synchronisation to be maintained. The radical processing involved in Header Removal is only one such “unacceptable” technique.
7.0.2.1.2. Bit-Identical Encoding Required
There is another reason why some forms of adaptation may be unacceptable to a mobile application. It is possible to produce other “lossy compression” schemes that might be appropriate for some traffic types. For example, HTTP (web) messages use a text encoding and could be re-encoded into a canonical form with compression. The resulting message would not be bit-identical. For most purposes, this is acceptable, but there are situations in which it is not; for example, if application-level integrity protection had been applied to the HTTP message, then this would fail when checked against the message that had been re-encoded to a canonical form.
Introducing a “Bit-Identical Encoding Required” flag could allow the PDCP entities to restrict their processing to adaptation that preserved the identical bit pattern of the message. Of course, it follows that indication of such a requirement would, by definition, mean that Header Removal was not allowed as this technique does not guarantee bit-identical transfer.
7.0.2.2. Traffic Type
It is proposed that a “Traffic Type” structure be introduced. This will indicate the traffic to be used within the associated bearer, and will include a parameter set the interpretation of which is specific to the Traffic Type carried. This is the Traffic Type Parameters.
For each different Traffic Type, the parameters might have a different structure or be empty. If an implementation receives such a structure and does not recognise the Traffic Type Identity value, it can ignore the whole structure, as this implies that it does not support a specific adaptation mechanism to process this traffic.
Traffic Type Identity:
(Unknown | IP | TCP | UDP | UDP/RTP | UDP/SIP | TCP/HTTP |,…)
The interpretation would be that the associated bearer is expected to carry packets of this type. There are several values that can be considered at this point; of these, only the RTP value is required for Header Removal to function. However, the others are given as potential examples; at present, all other values should be reserved.
• “Unknown” means that the kind of data carried in this bearer is completely unknown.
• “IP” means that the bearer is known to carry IP datagrams, but these hold a mix of TCP and UDP packets.
• “TCP” means that this bearer will carry TCP packets, but the kind of application level protocols carried in the TCP packets is unknown, or is a mix of protocols.
• “UDP” means that this bearer will carry UDP packets, but the kind of application level protocols carried in the UDP packets is unknown, or is a mix of protocols.
• “UDP/RTP” means that it is known that this bearer will carry only RTP packets.
• “UDP/SIP” means that this bearer will carry only SIP messages.
• “TCP/HTTP” means that the bearer will be used to carry web requests and responses only.
Traffic Type Parameters:
– Parameters (if any) associated with this traffic type
For Header Removal, the Traffic Type Identity ‘UDP/RTP’ is required. In this case, the Traffic Type Parameters will be interpreted as carrying Codec Type information. The internal structure of this sub-field is covered next.
7.0.2.2.1. Codec Type
Where the Traffic Type Identity is ‘UDP/RTP’, the associated Traffic Type Parameters should be interpreted as a list of triple values, each consisting of the Codec Identity, ACS Modes used, and the Payload Type associated with this Codec/ACS combination. It is valid for the length of the parameter to be zero (i.e for there to be an empty list of Codecs).
Conversely, note that there might be, in the future, more than one codec used for traffic carried in a single bearer, so the parameters for this traffic type should form a list of entries. For example, data reflecting DTMF-coded signals (encoded according to RFC 2833) might be interspersed with data from speech. The situations in which such use of more than one Codec Type is valid are for further study, but using a list structure does not preclude this possibility for future systems whilst ensuring “backward compatibility”. Each list entry consists of the following tuple:
Codec Identity:
(Unknown/Unspecified | GSM-FR | GSM-EFR | GSM-HR | AMR-NB | …)
Note – other values should be reserved.
ACS Modes Used:
Bit Set, with one entry per mode, each of which is a Boolean flag indicating whether or not this mode is part of the ACS. If the associated Codec does not use Active Codec Sets, then only one mode would be expected to be set true. The default value {00000000} (i.e. no modes in this set) should be used where modes are not known or are not applicable.
Note that the mapping between particular modes and positions in the bit set is TBD.
Also note that, to ensure forward compatibility, this bit field will need to hold flags for nine modes, to allow for the future introduction of AMR-WB to the GERAN.
Payload Type:
This is a copy of the Payload Type identifier to be used in RTP packets carrying data encoded according to the associated Codec Identity. This value is an 7 bit unsigned integer.
7.0.2.3. Example PASNAS Information
Combining these two field types, the following structure might be expected for the example described above, in which a bearer was to be used exclusively to carry RTP packets with a Codec Identity of ‘GSM-FR’, a Payload Type of 96, and for which the Mobile Application decided to state explicitly that it had no special requirements on the adaptation technique applied.
‘PASNAS Info’
{
‘Traffic Type’
Traffic Type Identity – ‘UDP/RTP’
Traffic Type Parameters – {{‘GSM-FR’, {00001000}, 96}}
‘Adaptation Type Requirements’
{SI= ‘false’, BiER= ‘false’}
}
7.0.3. Summary of the issues addressed in chapter 7
• How shall the SIP negotiation between the endpoints be performed, and how to make sure that the endpoints have all necessary information in order to complete the negotiation.
• The principle of how GERAN figures out which speech codec that has been selected in order to apply the appropriate channel coding schemes.
• The principle of how to signal/negotiate a change in codec during an ongoing call.
• The principle of how to select active codec set (ACS) when AMR is used.
• How and when header regeneration shall be applied.
• The principle of how GERAN figures out whether or not header removal may be applied.
• How the IP and port numbers are communicated between the UE and the PDCP entity in the BSS.
• How GERAN-GERAN, UTRAN-GERAN, GERAN-UTRAN handovers shall be performed with regard to header removal.
• How mid call SIP communication shall be performed.
•
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.1 Optimized voice call set-up within the IMS
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.1.1 Description of
|
Call set-up in the CS domain is based on the following principle:
1. The terminal announces its capabilities
2. The network select speech codec to be used
Call set-up in the IMS domain is based on a fundamentally different principle:
1. The terminal endpoints negotiate speech codec (or more generally media codecs) to be used
2. The terminal request the resources required, to the network.
The IMS SIP negotiation currently does problem not take into account any access specific information concerning the codec negotiation. This is particularly the case when the access network modifies the codec packets in some way as in header removal. The BTS may lack support for some of the channel coding schemes that corresponds to the speech codecs supported by the MS.
The solutions as proposed below may be combined. For example one solution can be adopted for initial implementation and may be further improved in combination with another solution.
7.1.2 Proposed solutions
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.1.2.1 MS knowledge of GERAN channel coding capabilities at the start of or before SIP negotiation
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.1.2.1.1 Description of the solution
|
This solution is based on the principle of letting the peer involved in the SIP call set-up know about the capabilities of the GSM/EDGE Radio Access Network (e.g. supported channel codings in the cell). Such knowledge has to be provided prior to, or during, the SIP-based call set-up. Several solutions are possible:
◦ The knowledge is provided as a new Information Element appended to the RADIO BEARER SETUP message, setting up the Radio Bearer for SIP signalling.
◦ Other solutions are possible and may be added
This solution is also based on the principle of having a deterministic rule for the BSS to work out that the RAB being established carries SIP-signalling. Several solutions are possible:
-
- Define a new Source Descriptor choice for SIP signalling - Make an on-demand request during the SIP negotiation (FFS).
- Other solutions are possible and may be added
When the user moves to another cell after SIP negotiation has started but before it is completed, the capabilities supported by GERAN may change. Several solutions are possible to handle this:
:
• The BSS handovers the resources used for the SIP Radio Bearer and the HANDOVER COMMAND or RB RE-CONFIGURATION message, whichever is used, can include such information for the new cell (see 44.018);
• The MS re-selects the new cell and sends a CELL UPDATE to the BSS. The response from the network can include such information for the new cell (CELL UPDATE CONFIRM or RB RE-CONFIGURATION).
• Other solutions are possible and may be added
If the channel coding capabilities supported by the old cell are not the same as those supported in the new cell, this may trigger codec re-negotiation at SIP level.
The impact on SIP level codec negotiation is then the following:
• In case of Mobile Originated call the selection of QoS attributes, codec, etc for each media flow described in the SDP contained in the SIP INVITE shall then take into account not only the SIP client own capabilities but also the capabilities of the GERAN. Each media flow will be associated to a list of all the codecs that are supported by both the originating SIP client and the controlling GERAN (as far as the necessary channel codings are concerned) and which fulfil the QoS required for the media flow. The SIP negotiation then takes place according to 3GPP TS 23.228.
• In case of Mobile Terminated call, when the addressed SIP client receives the SDP contained in the SIP INVITE, it shall then take into account the codecs that it accepts itself and that are supported by its controlling GERAN (as far as the necessary channel codings are concerned) before accepting the SDP and send the reply to the originating SIP client.
Such a solution will not require any SIP level codec renegotiation in cells where the same set of channel codings is supported by all transceivers. In case transceivers of a cell do not all support the same channel codings (e.g. some support TCH/FS and TCH/AFS codings, others support only TCH/FS), it may happen that a codec is negotiated at SIP level for which there is no transceiver availability at the time the Radio Bearer is set-up (e.g. AMR NB is chosen). This would imply SIP level codec renegotiation. This solution is therefore particularly suited for network deployments where a consistent set of channel codings is supported by all transceivers of a given cell. However, this does not require all cells of the network to support the same set of channel codings. This is further described in Annex A.
This solution may, if necessary, be further improved in combination with solution 7.1.2.3.
7.1.2.1.2 Pros and Cons- The SIP radio bearer is set up when the MS makes itself available to the IP Multimedia Subsystem. However, the SIP negotiation only takes place when a call is being received or initiated by the MS. Between these two events, a substantial amount of time may expire. During this time, the set of supported codecs may change due to high network load in the current cell, or because the user is moving into a new cell. This will lead to extra signalling between the MS and network.
7.1.2.2 SDP message delayed
7.1.2.2.1 Description of the solution
In this solution the proposal as described in 7.1.2.1 is enhanced. By delaying the final SDP message sent by the calling party until the resources have been allocated within the GERAN, there is no risk that a codec is selected that requires a channel coding scheme that is not supported in the BSS.
7.1.2.2.2 Pros and cons
• This solution will not work in the case where no mid path transcoding is carried out, such as in the case of IMS MS to IMS MS call where both mobiles are accessing the network via GERAN. The reason for this is that two different GERAN entities are involved in the SIP negotiation phase, and it has to be assumed that those GERANs may come up with different codec selections.
• This proposal changes the current working model for the IMS as defined in 23.228v5.0.0. This would cause substantial changes to the currently agreed information flows and would have to be agreed both in S2 and CN1. S2 has made a clear indication (LS Tdoc S2-011577) that:
“this solution should be removed from consideration”.
7.1.2.3
non-3GPP
7.1.3 Working assumption
Solution 7.1.2.1 is the current working assumption. Several sub-alternatives exist in 7.1.2.1, Nno agreement has been reached so far on working assumption on that level,
Solution 7.1.2.2 is removed from consideration.
Solution 7.1.2.3 requires quite some changes and additions in [6]. This make this solution non feasible in short term. However this solution add a value by outlining a future proof evolution of 7.1.2.1.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.2 Handling of ACS for AMR
|
In case of optimized speech with AMR codec there are additional issues that are related to managing the ACS as listed
7.2.1 Max four AMR modes can be part of an ACS I GERAN, at a time
7.2.1.1 Description of problemIn case of a session between GERAN MS and some other IP terminal, the IP terminal (somewhere in the IP cloud or in UTRAN) assumes that any of 8 modes are possible if the SIP level negotiation would result with AMR. However this is not true over the GERAN air interface as seen in R98 GSM AMR specifications. There could be maximum 4 modes
7.2.1.2 Description of solutions
7.2.1.2.1 AMR Format Parameters
This could be solved using MIME negotiation during the SIP/SDP where the ACS could be negotiated too. For example A party indicates (in SDP) ACS {12.2,7.95,7.4} and B party indicates ACS {10.2, 7.95,7.4}. So, the resulting common ACS would be {7.95,7.4}. It is clear that A party must only use modes included in the ACS that B party has indicated. Furthermore, although in general case the ACS means only the modes that a terminal is willing to receive, it seems quite clear that in GERAN case A party knows that it is only allowed to transmit modes included in its own ACS.
7.2.1.3 Working assumption
The MiME approach 7.2.1.2.1. is currently the GERAN working assumption
7.2.2 How to change the ACS at any given time
7.2.2.1 Description of problem
If we assume that only one codec and one ACS is agreed at the SIP negotiation, dynamic behavior of GERAN system (possibility to change ACS any time) would require SIP level re-negotiation of ACS. This re-negotiation is seen as incall modification of the session (SIP signaling during the speech call) and in order to transmit SIP signaling during the call, we have to use DTM like solution, so go to HR+HR constellation and this in turn requires changing ACS, since ACSs are different for FR and HR.
7.2.2.2.1.1 Description of solutions
7.2.2.2.1 use of a consistent Active Codec Set in geographic regions
In order to avoid SIP level negotiation a similar solution as described in 7.1.2.1, could be adopted. This would mean that a consistent set of ACS should be supported in the network.
7.2.2.3 working assumption
No working assumption has been reached so far
7.2.3 The encoder may have to use a more robust rate than the requested
Header removal functionality in PDCP will act as a proxy and receive AMR speech samples encapsulated in the RTP packet according to [6]. For downlink the speech samples are passed through channel encoder and the Mode Indication is set according to the information obtained from the AMR payload format for RTP. According to [6] the other end could ask using CMR (Codec Mode Request) field to receive a codec mode that would not be possible over the air interface in uplink at a certain time (or to be more precise it could be possible but the link quality could be so bad that the speech quality would be severely impacted). An example: The B party asks for 12.2, but the link conditions dictate the usage of more robust mode, for example 7.4. According to [6] GERAN PDPC header removal entity is mandated to send 12.2 in uplink, so it needs to set the Mode Command to 12.2 in the 2 AMR signalling bits. This issue is not unique and appears also in TFO cases. One simple solution would be to relax the requirement in [6].
Editors note: According to [6] this seems to already be possible. This section may be removed, or reformulated.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.2.4 How to force a change to an AMR rate able to be carried on a HR physical channel
|
7.2.4.1 Description of problem
There may bea need to change from Full Rate to Half Rate channels. This may be the case in high traffic load situations. It may also be necessary if DTM is used for SIP signalling and only one TS in UL and DL can be used (also refer to chapter 7.9).
7.2.4.2 Description of solutions
,
7.2.4.2.1 Choose a HR compliant ACS
One way to avoid SIP level re-negotiation is to choose an ACS that would be compliant with half Rate channels..In case of GMSK NB AMR this would mean to restrict the highest mode in ACS to 7.95. The implications of such restrictions should be evaluated.
Editors note: An example of signalling flow for MS initiated optimized speech is provided in appendix B.
7.2.4.3 Working assumptions
No working assumptions reached so far.
7.3 Radio Bearer Identification for GERAN
7.3.1 Description of problem
When GERAN is about to apply header removal, it is necessary for GERAN to identify which codec is used, as the corresponding channel coding algorithm has to be applied. Furthermore, in the case where AMR is used, GERAN must also be informed of which active codec set is used. GERAN can only handle up to four rates in its active codec set.
Editors note: The relation of operation of AMR over IP and GERAN’s limited active codec set needs to be clarified in cooperation with SA2.
7.3.2 Solutions
7.3.2.1 Direct communication between the UE and the BSC
7.3.2.1.1 Description of the solution
It is proposed to keep the exchange of information related to header removal completely within RRC. All required information is then transferred within extended RADIO BEARER SETUP messages as outlined in Figure 2.
Figure 2: Extended RB set-up procedure
1. The SGSN starts the setup of the RAB with RAB ASSIGNMENT REQUEST containing a generic QoS request as received from the UE via Session Management.
2. The BSC has no knowledge so far whether header removal could be applied to this RAB. Therefore the BSC will initiate the setup of a generic radio bearer according to the received QoS received in RAB ASSIGNMENT REQUEST. Within the RADIO BEARER SETUP message the BSC may include an indication to the MS, that header removal is supported in the RAN (e.g.: by sending a flag "Header Removal Supported"). Note that it is FFS whether the BSC shall indicate the support of HR at that point in time.
3. The MS has to check whether or not header removal is possible for that media stream. If this is the case, the MS sends all information needed to be able to apply header removal within a container inside the RADIO BEARER SETUP COMPLETE message to the BSC, i.e. a flag indicating "Header Removal Allowed", negotiated codec information and the RTP context.
4. The BSC detects that header removal, i.e. optimised voice can be applied. If the BSC decides to modify the (generic, not optimised) RB according to the information received from the MS in the RB setup complete message received, it starts a RB modification procedure by sending the RADIO BEARER RECONFIGURATION message to the MS.
If the BSC decides not to modify the RB it successfully terminates the RAB Assignment procedure instead of sending RB RECONFIGURATION to the MS.
5. MS sends back a RADIO BEARER RECONFIGURATION COMPLETE message.
6. BSC responds to SGSN with RAB ASSIGNMENT RESPONSE.
Now the establishment of the radio link is finished and a codec-specific channel coding will be applied. After this modified setup of the radio bearer, the signalling will be continued as described in Figure 2.
For MSs only supporting 1 TS in UL and 1 TS in DL the extended RB setup procedure might look different: the MS will not support the generic RB, because the amount of data for a generic RB will not fit into one timeslot. Therefore the MS has to reject the first radio bearer setup. But it could transfer all required information within the RADIO BEARER SETUP FAILURE message to the BSC and might also include the flag “Header Removal Allowed”, which indicates to the BSC that the setup of a RB with applied header removal will be successful.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.3.2.1.2 Pros and cons
|
+ no impact to CN
+ if header removal is not to be applied to the media stream, the RB setup procedure remains unchanged (except for the transfer of the “Header Removal Supported” flag from the BSC to the MS and the “Header Removal Allowed” flag set to false from the MS back to the BSC).
• Two more messages will be required to setup the RB for optimised speech. However the significance of the added delay within the whole setup procedure has to be verified.
• It is unclear how a CDR shall be generated in order to be able to charge differently for optimized voice and voice carried over a generic radio bearer
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.3.2.2 SDU format information approach
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.3.2.2.1 Description of the solution
|
Detailed QoS information is provided in the ‘Activate PDP context request’ message by using the ‘SDU format information’ attribute. This information uniquely identifies the appropriate channel coding in the GERAN. However, ‘SDU format information’ would have to be introduced in R5.
For multi rate codecs such as AMR, it is important that the SDU format is provided for all rates even though only a subset has been negotiated on SIP-level, in order for GERAN to be able to identify the codec unambiguously.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.3.2.2.2 Pros and cons
|
- The solution proposed does not specify how a potential future codec is uniquely identified if that codec has exactly the same bit mapping and protection for each class of bits in the payload format of an existing codec.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.3.2.3 Activate PDP context request message approach
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.3.2.3.1 Description of the solution
|
Following the SIP negotiation, which needs to result in one desired codec, the UE expresses this request explicitly by stating the desired codec in the subsequent resource request to the network. A field containing the specific speech codec desired is introduced in the ‘Activate PDP context request message’ to the SGSN, by extending the QoS information element. More specifically, the codec information can be an extension of the ‘Source Statistics Descriptor’ field that will be part of the QoS IE in R5. (The R99 QoS information element included in the Activate PDP context request message is shown in section 7.5.2.).
This information is then passed to the GERAN at the ‘Radio Access Bearer Request’, by also extending the ‘Source statistics descriptor’ in the RAB QoS parameter set.
For AMR, it is assumed that the preceding SIP negotiation not only results in ‘AMR’, but rather AMR plus a preferred active codec set consisting of four or less rates. This active codec set information is then conveyed from the UE to GERAN. Thus, in case of AMR, the new field in the QoS information element, sent from the UE via SGSN to GERAN, comprises both AMR and the preferred active codec set.
Editors note: This section may be updated to reflect concerns expressed on service specificity. It is intended to place the codec information within a transparent container to be relayed via the SGSN.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.3.2.3.2 Pros and cons
|
+ This solution is straightforward and imposes limited changes to existing standards. It is architecturally clean in that it uses existing messages for resource requests from the UE to GERAN. The codec information can potentially be used by other purposes as well, for example charging.
- Its potential drawback is that the PDP context message, which is a request for a bearer service, includes application-related information. To avoid this, one could consider the ‘SDU format information’ approach (section 7.3.2.2), which however introduces a bigger impact on the PDP context message size.
7.3.3 Working assumption
Currently option 7.3.2.3 seems to be the most promising solution.
7.4 Limitations due to RTP handling
Editors note: This section is to be restructured.
The Sequence Number (SEQ) and Timestamp (TS) in the RTP header determine the time instant when the contents of a packet is played out at the receiver. The SEQ is expected to increment by one at the receiver, otherwise it will be interpreted as a gap in the sequence. Also, the first TS value received is used as a reference at the receiver. This reference together with a timestamp determines the presentation time of subsequent RTP packets. During handover events (and possibly during normal operation), positive or negative slips in sequence numbers may occur. Depending on the size of the slip this may cause degradation of speech quality. A positive drift in a subsequent timestamps will cause the RTP receiver to generate a silence period. The length of this silence period will be equal to the drift in seconds. A negative drift in the timestamp will cause the RTP receiver to drop the packet, since from its perspective, the presentation time for the contents of the current packet has passed.
7.5 Identification of header removal allowed
7.5.1 Description of problem
As described in chapter 7.3, GERAN will be made aware if a supported speech codec is used, and if so, which one. However, it is also necessary for GERAN to identify whether or not it is allowed to use header removal. If the speech media flow is part of a multimedia application requiring synchronisation of the different media flows, header removal is not allowed.
7.5.2 Solutions
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.5.2.1 Activate PDP context request message approach
|
7.5.2.1.1 Description of the solution
This solution is based on the principle of providing the information whether or not header removal is allowed in the Activate PDP context request message. Several solutions have been presented how to name these bits:
1. ‘Header removal allowed’ bit.
2. ‘This flow may be synchronized with other flows’ bit. (Synchronization Indicator from PASNAS)
3. The bit may be part of a transparent container delivered to the GERAN via the SGSN.
7.5.2.1.1.1 Header removal allowed bit Since header adaptation mechanism is dependent on the application (e.g. in case of VoIP only application header removal is possible) one solution is that the MS indicates the header adaptation mechanism to be applied for a particular PDP context. The indication could be part of the Quality of Service IE, and thus the solution can be combined with the solution presented in section 7.3.2, solving also the radio bearer identification problem.
The signalling flow for the solution is given in the figure below:
The application will use the SIP signalling for setting up the session, and UE is the entity that knows the type of application used for the session.
After the initial phase of SIP signalling is completed (i.e. the session description has been agreed), the UE will activate the PDP context. Specifically in case of optimized speech (VoIP with header removal) the UE will send the Activate Secondary PDP Context Request message to the network. This message contains the Quality of Service Information Element. New field is needed in QoS IE to indicate the preference of the header adaptation mechanism for the particular PDP context. An example of the field could be as shown in the following table. Table shows the QoS IE as specified in 24.008 v4.1.1.
8
7
6
5
4
3
2
1
Quality of service IEI
octet 1
Length of quality of service IE
Octet 2
0 0
spare
Delay
class
Reliability
class
octet 3
Peak
throughput
0
spare
Precedence
class
octet 4
0 0 0
spare
Mean
throughput
octet 5
Traffic Class
Delivery order
Delivery of erroneous SDU
Octet 6
Maximum SDU size
Octet 7
Maximum bit rate for uplink
Octet 8
Maximum bit rate for downlink
Octet 9
Residual BER
SDU error ratio
Octet 10
Transfer delay
Traffic Handling priority
Octet 11
Guaranteed bit rate for uplink
Octet 12
Guaranteed bit rate for downlink
Octet 13
Spare
Header Adaptation
Octet 14
Figure 10.5.138/TS 24.008: Quality of service information element
Table 10.5.156/TS 24.008: Quality of service information element
Header Adaptation (Octet 14)
Bits
2 1
In MS to network direction:
0 0 No header Adaptation preferred
0 1 Header Removal preferred
1 0 Header Removal not possible
1 1 Spare
The SGSN send the RAB assignment request as specified in 25.413 and include the proposed "Header Adaptation" field in RAB Parameters IE. SGSN could as well use predefined QoS parameter combination in the RAB assignment message which would give unambiguous information to GERAN that header removal can be used.
When receiving the RAB assignment request, radio access network would choose the header adaptation mechanism according to its algorithm and inform the UE using Radio Bearer Set-up message.
The example shown above is only one possibility on how to convey the necessary information to the radio access network. If this solution is combined with the solution described in section 7.3.2.3 (dealing with the problem of radio bearer identification), there is potential room for parameter optimisation. One possible scheme is that an explicit codec indication (according to 7.3.2.3) by default implies that header removal is allowed and preferred, making a specific ‘header adaptation’ field superfluous. Such syntax details are FFS.
7.5.2.1.1.2 ‘This flow may be synchronized with other flows’ bit
Editors note: This is an alternative way of providing the necessary information (as described in chapter 7.5.1) from the terminal to the GERAN. This section is to be completed.
7.5.2.1.1.3 Information provided in a transparent container
Editors note: This is an alternative way of providing the necessary information (as described in chapter 7.5.1) from the terminal to the GERAN. This section is to be completed.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.5.2.1.2 Pros and cons
|
+ This solution has the advantage that it implies very limited changes to existing specifications.
- A possible drawback is that that higher protocol messages such as the PDP context messages have to convey header adaptation information, which can be considered as being radio access related. Given the nature of optimized speech and its relation to the application setup, this drawback would seem inevitable.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.5.2.2 Direct communication between the UE and the BSC
|
A different method is used to indicate that HR is possible when using the direct communications approach (see section 7.3.2.1.1).
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.5.3 Working assumption
|
No agreement reached so far.
7.6 IP and port number information transfer from MS to GERAN
7.6.1 Description of problem
In order to carry out header regeneration in the uplink the relevant information must be communicated with the PDCP entity in the GERAN. A number of possibilities have been identified, so far, in order to transfer IP and port numbers from the MS to PDCP in BSS.
7.6.2 Solutions
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.6.2.1 RRC signalling approach
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.6.2.1.1 Description of the solution
|
The information is provided by RRC signalling at RB set-up.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.6.2.1.2 Pros and cons
|
Editors note: To be completed.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.6.2.2 TFT approach
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.6.2.2.1 Description of the solution
|
The information is sent in a TFT from the MS to SGSN, which in turn provides the information to the BSC.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.6.2.2.2 Pros and cons
|
Editors note: To be completed.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.6.2.3 Direct communication between the UE and the BSC
|
A slight variation on the RRC signalling method is used when using the direct communications approach (see section 7.3.2.1.1). The data is included within the message sent by the MS during the extended RB setup procedure.
7.6.3 Working assumption
Currently solution 7.6.2.1 seems to be the most promising solution. However the expertise of TSG RAN and TSG SA is needed in order to make a decision.
7.7 Handover issues in optimized voice
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.7.1 Description of problem
|
When inter BSS, inter RAN or BSS-RAN handover takes place, the header generation context may have to be relocated. A mechanism for this purpose is needed. In addition, it should be clarified how slips in RTP sequence numbers and timestamps can be minimized or completely eliminated.
7.7.2 Proposed solutions
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.7.2.1 Time stamp and sequence number handling during a handover
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.7.2.1.1 Description of the solution
|
This solution assumes that handover is carried out as specified in 44.018 and that relocation follows the procedures that have been specified in 25.413 and 23.060. As a part of the relocation of the RNS context the location of the header removal / generation function is moved from the source BSS to the target BSS.
In the case of GERAN to GERAN handover, a way to ensure smooth continuation of the time stamp value is to utilize synchronized clocks in the network entities carrying out header removal/generation. This has been illustrated in figure 3. The MS sends voice frames 1-4 via source GERAN. Header generation function creates RTP packets and uses local clock to generate the time stamp information for each packet. After sending the relocation commit and handover command messages the "data path" is switched to go via target RAN. The clock synchronization is utilized by including the latest time stamp information and the corresponding clock time in the Relocation Commit (or Forward SRNS Context) message. When the target RAN receives the message it can, based on the local clock and the received information, deduce the right time stamp value. Some frames may be lost during the handover but that should not cause any problems as long as the time stamp value continues without disruption.
Editors's note: In here clock synchronization does not mean BTS synchronization but merely that the clocks in network entities carrying out header generation have been setup to the same time and are reasonably close to each other in rate.
Figure 3. Time stamp synchronization in GERAN–GERAN handover (Note that in case of AMR for each 20ms frame time stamp increases by 160).
In case of GERAN to UTRAN handover the header adaptation mechanism changes from header removal to header compression and the location of the RTP end point moves from the network to the terminal. In this case large jumps in the field values are avoided by transferring the time stamp, sequence number fields and the TDMA frame number from the network to the terminal inside a container in the Handover To UTRAN Command.
When the MS receives the handover command it can deduce the correct time stamp value from the current TDMA frame number and the received information. The procedure is illustrated in figure 4. In the example the RTP packets 1-3 are sent through GERAN using header removal/generation. After sending the third RTP packet the network sends a handover command to the terminal containing the TDMA frame number when the packet 3 was sent and the corresponding time stamp and sequence number information. In the terminal the right RTP time stamp value can be deduced from the received information.
Figure 4 Time stamp synchronization in GERAN – UTRAN HO / relocation.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.7.2.2 Pros and cons
|
- The proposed solution may lead to small drift in the transferred field values. It is the assumption that this does not cause large quality degradation. However, this needs to be verified from IETF AVT group. The size of the drift will be directly proportional to the number of muted or discarded frames as explained in section 7.4
7.7.3 Working assumption
No agreement reached so far.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.8 Mid call legacy codec support
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.8.1 Description of problem
|
The Radio Access Network infrastructure may not support all possible channel coding schemes in all areas, and, potentially, the set of channel coding schemes supported in one area may be completely different from the set supported elsewhere. If an IMS call is active and uses Header Removal (and so relies on an unequal error protected channel coding scheme associated with the current CoDec), this can cause problems in mid-call.
7.8.2 Solutions
If, during a call, a resource that has been used is no longer available, there are two choices to resolve this problem.
Either:
• The PDCP Mode must be changed from Header Removal to Header Compression (and the radio bearer should be configured to use an equal error protected channel coding scheme), or
• The Codec used in the media stream will need to be changed to one that is associated with a supported unequal error protected channel coding scheme
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.8.2.1 PDCP Mode Change
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.8.2.1.1 Description of the solution
|
Editors note: To be completed
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.8.2.1.2 Pros and cons
|
Editors note: To be completed
7.8.2.2 Mid Call Codec Change
It is assumed that the call control entities must maintain a valid specification of the media transport in use.
If the codec used is to change in mid-call to one not specified in the existing session description, then the description agreed by the SIP end points at the start of the call will no longer reflect the actual media streams being exchanged. From the above assumption, this will require SIP messages to be exchanged "end to end" holding a replacement session description. This is shown in section 7.8.2.2.1 – 1.
If the codec change is to one included already in the existing session description, then alternatives not requiring SIP message exchanges may be used; these are covered in section 7.8.2.2.1-2.
Note that, if the session description includes only one codec at the end of call setup, then there is no alternative to engaging in a SIP call re-negotiation. The “non-SIP” alternatives assume that there is more than one codec included in the session description at the end of call setup.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.8.2.2.1 Description of the solutions
|
1. SIP call re-negotiation
[Standard IMS procedure as will be described in TS24.228]
2. Non-SIP Codec change signalling
If a media description, at the end of call set up phase, includes a set of alternative CoDecs with more than one member, then a change in CoDec between these listed alternatives would not invalidate the session description agreed during call setup, and so no SIP message exchanges would be needed in this event.
It is assumed that listing more than one alternative within the session description does not negate the requirement that the same codec be used in both directions of a call at any one time. Although, in principle, such a session description might seem to allow different CoDecs to be used in either direction, the policy will be to only support the bi-directional case. To maintain this policy, any change to the codec used by an end point should be signalled to ensure that both end points change codec at the same time; an end point should not simply decide to swap CoDecs without agreeing this with its peer.
There are several options for signalling a codec change without the use of SIP message exchanges. These are covered next.
a. RTCP Message Exchange
This approach is based on exchanging RTCP messages between the RAN that detects a resource problem and the remote system, using the “fast feedback” scheme. It has two variants; one variant proposes to use Sender Report and Receiver Report messages to carry indications between the network-based PDCP entities of a proposed codec change. The other variant uses the “Application-specific” message type to carry the indications between the peer entities This solution proposes to use RTCP to change/re-negotiate the ACS during an RTP session. The RTP proxy or in header removal scenario the header removal/generation function would send RTCP packets containing information regarding the allowed codec modes (ACS) whenever the allowed codec modes changes. The terminal would not participate in this signalling at all because it is the GERAN who decides the ACS. The RTCP packets should not be sent over the air interface.
RTP/RTCP protocols provide two alternatives to realize this: In addition to 'regular' RTCP Sender Reports (SR) and Receiver Reports (RR), it is possible to extend the RTCP functionality with application/payload type specific feedback messages. There seems to be two mechanisms to extend RTCP to support the idea presented here:
1. Section 6.4.3 in [3] specifies a possibility to define an extension field to RTCP SR or RR.
2. Section 6.7 in [3] specifies a possibility to define an application specific RTCP packet type.
There is a work in progress in IETF AVT group on 1, see [9],[10], and it seems like a suitable mechanism to convey AMR ACS update during a session.
Higher level protocols are added on top of RTCP and RTP to allow advance indication (and negotiation) of codec/Payload Type changes. Any such scheme must provide its own reliability mechanism as RTCP and RTP are unreliable protocols.
Editors note: An example of such higher-level protocol is outlined in G2-010020. This particular solution describes an RTP-based solution.
Editors note: The backup solution for the case when the scheme is not supported is an abrupt codec change, resulting in transient packet loss greater than if advance notice would have been given.
Editors note: A procedure for layer 3 messaging between the BSC and MS is required when a new ACS (or codec) has been agreed using RTCP or RTP signalling. This is FFS.
“In Band” Signalling
This approach works by injecting RTP packets into the existing media stream sent towards the core network, and detecting RTP packets that have been injected by the remote peer.
[For Details, see contribution G2-010020]
Editors note: The backup solution for the case when the scheme is not supported is an abrupt codec change, resulting in transient packet loss greater than if advance notice would have been given.
7.8.2.2.2 Pros and Cons
• Although using SIP signalling would appear to be the simplest solution, it does have some problems. First, it requires call control signalling to be carried over the air interface. Secondly, it is not easy to see how the Terminal can be informed that it should engage in SIP message exchanges during a Handover; although the GERAN detects the resource problem, it is not a party to call control signalling and so it must have some way to instruct the Terminal to carry out these exchanges. Such an approach would require the expertise of SA WG2 and CN WG 1 groups to clarify the appropriate procedures.
• Both the non-SIP approaches have one major benefit; they do not need any extra signalling to be carried over the air interface (over and above the necessary radio bearer modification procedures that are required on any change to the bearer). Both require a specialised application protocol to be used on top of the existing RTCP or RTP transport protocols. Of the two, the RTCP-based approach would seem to require an extra PDP context to be arranged; how this is done by the BSC is unclear. In addition, this approach has raised some other concerns; it is questionable if it is wise to generate RTCP SR/RRs when the RTP protocol is terminated in the MS and RTCP is terminated in the BSS. In such an architecture, the RTCP RR will contain information about quality in the BSS, not in the MS. It is suggested that it may not be appropriate to make use of RTCP SR/RR if the termination point of the RTP protocol is not in the same node as the RTCP protocol.
• If no RTCP SR/RRs are generated (for the above mentioned reasons), then with the other variant (using “Application-specific” messages), RTCP would be used for the sole purpose of providing a possibility of informing the BSS of a change in the codec or ACS.
• Furthermore, the usage of RTCP for this task is questioned, since RTCP is not a reliable signalling protocol. There is no way of ascertaining that the ACS change has been received correctly, so that more details are required on the way in which the end points can exchange application level indications reliably.
• The RTP based approach does not have the problems of the other schemes, but (in common with the RTCP-based approach) does require that the alternatives are included in the “final” session description agreed at call setup. This solution assumes that it is allowed to negotiate multiple codecs for a SIP-session. Whether this is the case is FFS.
It cannot be ensured that the special RTCP or RTP functionality is deployed in all conceivable endpoints (also non-3GPP).
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.8.3 Working assumption
|
The current working assumption is based on SIP codec renegotiation 7.8.2.2.1-1. All other schemes as described in this chapter are FFS.
7.9 Bearer support for mid call SIP signalling
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.9.1 Description of problem
|
It is foreseen that there may be additional mid call IMS SIP communication using header removal.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.9.2 Solutions
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.9.2.1 Solution A
| |
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.9.2.1.1 Description of the solution
|
The following means can be used for SIP signalling:
3. FACCH
4. Downgrade to HR channel. This requires further analysis of:
a. TBF allocations for signalling
b. The codec selected at the SIP negotiation must be able to be reconfigured to support a HR channel, without SIP level renegotiation.
5. Allocation of additional timeslot
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.9.2.1.2 Pros and cons
|
Editors note: To be completed
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
7.9.3 Working assumption
|
Solution 7.9.2.1 has been accepted as working assumption.
8 Header compression in GERAN
Editors note: To be completed
9 Recommended work for GERAN voice optimization schemes
Editors note: To be completed.
9.1 Recommended work for particular groups
Editors note: To be completed.
|
1549644dc078ef9fd6712bc573d8101d
|
43.900
|
10 Open issues
|
This section lists identified open issues related to support of voice optimisation for the IMS in the GERAN.
Ref
Description of problem
Status
1
Is it necessary to define one channel coding scheme as mandatory in the standard, required to be supported in all GERAN based IMS SIP based calls?
2
Is there a requirement for the operators to prioritise other channel coding schemes than the default channel coding schemes to be used in the SIP negotiation?
3
In the case of a IMS user in a communication exchange to a non SIP user where a signalling translator is needed on the control plane to translate SIP messages to the call control used by the other party.
1. What is the status regarding this work in CN3?
2. Is the signalling transition transparent to the end systems?
LS sent to TSG CN3
(TDOC: OVS-01043)
4
What information is available to the UTRAN/GERAN applicable to uniquely identify IMS signalling SIP messages to enable specific treatment by the GERAN (i.e. sending the SIP message over the FACCH).
1. What is the current status and development of the SIP signalling work?
2. Can the QoS (RAB) parameters, so far defined for RANAP be utilised to distinguish RABs for SIP messages (signalling) targeted to/from the proxy CSCF from RABs for the actual speech transport?
3. Can the QoS (RAB) parameters, so far defined for RANAP be utilised to distinguish RABs for SIP messages (signalling) not targeted to/from the P-CSCF (non IMS Signalling) from RABs for the actual speech transport?
4. Can the QoS (RAB) parameters, so far defined for RANAP be utilised to distinguish RABs for SIP messages (signalling) not targeted to/from the P-CSCF (i.e. non IMS signalling) from those targeted to/from P-CSCF?
5. How are the QoS assignments attributed to the SIP signalling in each of the cases stated above?
LS sent to TSG SA2 and TSG RAN3
(TDOC: OVS-01044))
5
In the Optimised Voice service within GERAN, only one codec (and if applicable the AMR Active Codec Set (ACS)) will be the consequence of the SIP negotiation. Is the resulting single codec decision an IMS restriction?
LS sent to TSG SA2, TSG CN1 and TSG SA4
(TDOC: OVS-01046))
6
In the Optimised Voice service using AMR, an indication of the ACS has to be made at the SIP negotiation level. In GERAN a set of four or less rates has to be selected within the AMR codec. The current solution being discussed within GERAN is that the negotiation of ACS on a SIP level is done using MIME encoding of format parameters.
In order for this mechanism to work, all entities must understand the request. This raises the issue that the MIME encoding would need to be included in the standards as a mandatory requirement. Is this ability currently defined in the standards?
LS sent to TSG SA2, TSG CN1 and TSG SA4
(TDOC: OVS-01046))
7
If AMR is used; is there a mechanisms that can enforce the use of an AMR mode that can be carried on a physical HR channel (i.e. AMR 795 or lower) within the RTP for carrying Optimised Voice in GERAN?
LS sent to TSG SA2, TSG CN1 and TSG SA4
(TDOC: OVS-01046))
8
GERAN is currently looking into the analysis of the different mechanisms it can use for carrying mid call SIP messages over the GERAN.
Which are the typical mid call SIP signalling scenarios considered in CN1 call flows?
LS sent to TSG SA2, TSG CN1 and TSG SA4
(TDOC: OVS-01046))
9
What is the current status on PDCP context transfer at handover in RAN2 and when is it planned to be completed?
Editors note: This section has been agreed on a general level at the joint GERAN/SA2 meeting in Helsinki (1-3 of August 2001). However as it was added after the meeting, all companies have not had the opportunity to review it in detail.
Annex A:
Dimensioning principles
It is assumed that legacy transceivers will not be able to support all future channel-coding schemes. The concepts as described below will allow minimizing the number of mid call codec changes. Such codec changes may involve SIP signalling, which may be very detrimental to the perceived voice quality.
A.1 The buffer zone concept
Channel coding capabilities for new codecs may be launched and introduced in limited but homogenous geographical areas. Thus when upgrading the network, all cells in a given area are updated to support the new channel coding schemes. If this is done before the operator allows the speech codec associated with this new channel coding scheme to be used for call set-up, then the number of mid call codec changes will be minimized.
Figure X. Mid call codec renegotiation will only have to take place if a call is set up in a cell marked (0) and the customer moves into some of the cells marked (1).
A.2 The layering concept
Considering a layered cell planning, channel coding capabilities for new codecs may be launched and introduced in one or more layers of the network but not all of them simultaneously. Thus when upgrading the network, all cells in a given layer are updated to support the new channel coding schemes. At call set-up the network can direct the MS to one layer, depending on the MS capabilities. By ensuring that the MS will remain in that layer for the duration of the SIP session (e.g. forbidding handovers between different layers, or at least between layers that do not support the same set of channel coding capabilities) then the number of mid call codec changes will be minimized.
Figure X. Mid call codec renegotiation will only have to take place if resources for a call are handed over from layer 0 to layer 2, or from layer 0 to layer 1 and there are no resources available for codec 1.
A.3 Resource dimensioning concept
Cells may be dimensioned by the operator, in such a way that a sufficient number of channel coder resources (non legacy transceivers) are available in each cell, in the areas where a certain channel coding scheme is used. This has the implication that the operator will dimension the channel coder resources in response to congestion detected in the cell.
Mid call codec change will happen in the case where no appropriate channel coding resources can be allocated. This can occur in two cases. Either as a result of resource exhaustion locally or in the remote RAN.
Annex B:
An example of signalling flow for MS initiated optimized speech
The following is a simplified example of signalling for optimized speech taking into account some potential solutions described so far in the TR.
Figure X. Example of signaling flow for optimized speech
Annex C:
Change history
Change history
Date
TSG #
TSG Doc.
CR
Rev
Subject/Comment
Old
New
Annex D: change history up to v. 1.0.0
(to be deleted when the TR reaches version 2.0.0)
Version
Date
Comment
0.0.1
June 01
Created by GERAN 2 #5bis
0.0.2
8 Aug. 01
Output from the SA2 & GERAN Joint Meeting (RAN delegates invited) on IMS issues and optimised voice services
1.0.0
August 01
Stopped at GERAN#6
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
1 Scope
|
This technical report will propose an architecture that provisions an all-IP architecture option for release 00. The purpose of the technical report is to
• identify key issues and affected ongoing 3GPP work that need to be resolved and
• propose a high level work plan for completion of an all IP release 00 UMTS standard
in order to provide this architectural option within Release 2000.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
2 References
|
The following documents contain provisions which, through reference in this text, constitute provisions of the present document.
• References are either specific (identified by date of publication, edition number, version number, etc.) or non‑specific.
• For a specific reference, subsequent revisions do not apply.
• For a non-specific reference, the latest version applies.
• A non-specific reference to an ETS shall also be taken to refer to later versions published as an EN with the same number.
[1] TS 22.101 version 3.6.0: Service Principles
[2] TS 23.121: Architectural Requirements for Release 99.
[3] TS 22.121: The Virtual Home Environment
[4] TS 23.002: Network architecture
[5] "Compressing TCP/IP Headers for Low-Speed Serial Links", IETF RFC 1144, V. Jacobson
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
3 Definitions and abbreviations
| |
8a9283a656db20cd080be5c53231664c
|
23.922
|
3.1 Definitions
|
For the purposes of the present document, the [following] terms and definitions [given in ... and the following] apply.
existing service: services supported in Release 99 and earlier releases for both GSM and UMTS.
All IP core network: core network of release 2000 that uses IP for transport of all user data and signalling
ERAN is defined as an evolved GSM BSS supporting EDGE modulation schemes on a 200kHz basis and real time packet services
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
3.2 Abbreviations
|
For the purposes of the present document, the following abbreviations apply:
<ACRONYM> <Explanation>
2G second generation
3G third generation
AMR Adaptive Multi Rate
AS Application Server
BSC Base Station Controller
BTS Base Station
CAMEL Customised Applications for Mobile Network Enhanced Logic
CAP CAMEL Application Part
CC Call Control
CCF Call Control Function
CN Core Network
CS Circuit Switched
CSCF Call State Control Function
CSE CAMEL Service Environment
DN Directory Number
DNS Directory Name Server
EDGE Enhanced Data for GSM
EGPRS Enhanced GPRS
FFS For Further Study
GGSN Gateway GPRS Support Node
GMSC Gateway MSC
GPRS General Packet Radio Service
GSN GPRS Support Node
GTP GPRS Tunnelling Protocol
H-CSCF Home CSCF
HN Home Network
HSS Home Subscriber Server
ICGW Incoming Call Gateway
IN Intelligent Network
INAP IN Application Part
IP Internet Protocol
ISDN Integrated Services Digital Network
ISP Internet Service Provider
ISUP ISDN User Part
LAN Local Area Network
LN Logical Name
MAHO Mobile Assisted Handover
MAP Mobile Application Part
MCU Media Control Unit
MExE Mobile Execution Environment
MGCF Media Gateway Control Function
MGW Media Gateway
MM Mobility Management
MO Mobile Originated
MRF Media Resource Function
MSC Mobile Switching Centre
MT Mobile Terminated/Terminal
NPA Numbering Plan Area
O&M Operations and Maintenance
ODB Operator Determined Barring
OSA Open Service Architecture
PCU Packet Control Unit
PDP Packet Data Protocol
PDU Packet Data Unit
PS Packet Switched
PSTN Public Switched Telephony Network
QoS Quality of Service
R00 Release 2000
R99 Release 1999
RA Routing Area
RAN Radio Access Network
RLC/MAC Radio Link Control/Media Access Control
RNC Radio Network Controller
R-SGW Roaming Signalling Gateway
RTP Real Time Protocol
SAT SIM Application Toolkit
SCF Service Control Function (IN) and Service Capability Features (VHE/OSA)
SCP Service Control Point
S-CSCF Serving CSCF
SGSN Serving GPRS Support Node
SIP Session Initiated Protocol
SLA Service Level Agreement
SN Serving Network
SRNC Serving Radio Network Controller
SSF Service Switching Function
TCP Transmission Control Protocol
TE Terminal Equipment
T-SGW Transport Signalling Gateway
UDP User Datagram Protocol
UE User Equipment
UMS User Mobility Server
UTRAN UMTS Terrestrial Radio Access Network
VHE Virtual Home Environment
VLR Visitor Location Register
VoIP Voice over IP
VPN Virtual Private Network
WAP Wireless Application Protocol
WIN Wireless IN (ANSI-41)
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4 Requirements
|
In order for TSG-SA2 to conduct a study of the architecture issues relating to the introduction of an All IP architecture within UMTS, assumptions were made as to the requirements for this architecture. TSG-SA1 are invited to validate and to extend these requirements as part of the work package on requirements for Release 2000.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.1 General
|
The aim of the all IP architecture is to allow operators to deploy IP technology to deliver 3rd Generation services, that is an architecture based on packet technologies and IP telephony for simultaneous real time and non real time services. This architecture should be based on an evolution from Release 99 specifications and should be compatible with IMT-2000, providing global terminal mobility (roaming) [1].
The IP network should provide wireless mobility access based on ERAN and UTRAN with a common core network, based an evolution of GPRS, for both. In this context, an E-GPRS radio access network is a 200kHz GSM based network supporting EDGE and evolved to support real time packet services. Although EDGE is not within the scope of 3GPP, there are requirements for the core network of the all IP architecture, to be common to both access technologies.
The characteristics of this network are
• Based on an evolution of GPRS
• Common network elements for multiple access types including UTRAN and ERAN
• Packet transport using IP protocols
• IP Client enabled terminals
• Support for voice, data, real time multimedia, and services with the same network elements.
The report also covers the support of CS services on IP technologies.
The benefits of this approach include
• Ability to offer seamless services, through the use of IP, regardless of means of access (e.g. common features used by subscribers whether accessing via conventional land telephony, cable, wireless, HIPERLAN 2 etc.)
• Synergy with generic IP developments and reduced cost of service
• Efficient solution for simultaneous multi-media services including voice, data, and advanced real time services.
• Higher level of control of services
• Integrated, and cost reduced OA&M through IP
• Take advantage of Internet applications by supporting terminals which are IP clients.
• Cost reduction through packet transport
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.1.1 General Requirements
|
1. The overall aim of the all IP network is to support similar services to GSM release ’99 and new innovative services. Where appropriate these services should inter-work with exisiting GSM services.
2. In addition it should also possible to support existing (R99 and before) services/capabilities (speech, data, multimedia, SMS, supplementary services, VHE,...)in a manner that is transparent to the users of these services [1]. That is, the network needs to provide the service capabilities required in such a way as to support interworking of these services between the R00 all IP network option and the other family networks two domain architecture option (GSM pre Release 99, UMTS release 99).
3. The standard shall enable the all IP core network to support release 99 CS terminals. This shall be standardised in such a way as to allow operators to decide whether or not they wish to support Release 99 CS only terminals.
4. The support of existing services shall not preclude the extension of service capabilities possible through the use of an all IP architecture.
5. When the all IP networks are deployed, there will be services and databases provided for existing networks which are non-IP based e.g. local number portability, free phone numbers, specialised corporate services. The all IP architecture will need to be able to access these services.
6. R’00 all IP core network shall allow implementations having a CS and a PS domain, that are separated like both these domains in the R’99 architecture. This implementation allows the two domains to evolve independently, e.g. to combine an all IP R’00 PS domain with a STM based R’99 CS domain. Furthermore it shall be possible to implement a CS domain that uses all IP based architecture and in distinct service areas of the same network a CS domain based on ATM/STM. This allows a smooth migration to an all IP based core network.
The R’00 all IP architecture shall support that all services share bearer level transport and bearer control.
R’00 architecture shall allow an operator to migrate a R’99 network into a R’00 network, without need for change of transport network technology, node numbering scheme etc. R’00 networks shall also allow connection of R’99 UTRAN over Iucs, to provide the operator with flexibility in the network implementation.
Note: In the general R’00 architecture other transport technology than IP shall be possible.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.2 Service Capabilities
| |
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.2.1 General
|
The following general service capabilities are identified:
1. Legal interception has to be possible in the R00 All-IP network.
2. Emergency calls shall be supported in the R00 All-IP network.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.2.2 Basic requirements for the Service and Application Platforms
|
List of requirements on the “Application and Service” block:
1. Service capabilities are to be made available to the Applications through Service capability features;
2. Service capability features are provided by one or more service capabilities (possibly directly provided by network functions, e.g. HSS, CSCF etc.), as illustrated in Figure 4-1;
Figure 4-1: Relationship between Service Capability Features (SCFs) and Service Capabilities.
3. Incremental introduction of service capability features by the network operator has to be possible, for operator specific service capabilities;
4. A standard interface, called the application interface, has to be introduced for access to Service capability features by Applications. This interface has to provide a controlled, secure and accountable relationship between Applications and Service capability features;
5. The Application interface must provide access to the Service capability features based on user subscription profile;
6. Applications can be located either on servers and/or on (mobile) terminals;
7. Applications can only access the service capabilities through the service capability features;
8. Applications can utilize both capabilities provided by the Mobile Network functions, and functions as provided by IT systems, through the service capability features.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.3 Numbering Schemes
|
The standards shall allow mobile terminated communications to be routed to the user’s terminal on the basis of a single identifier e.g. MSISDN. This does not preclude multiple addresses being used for different services and capabilities (e.g. data, Fax, SMS). The network will route the call to the terminal over the available resources, dependent upon, for example, terminal capability, traffic loading and coverage. Networks migrating to an all IP architecture will require the ability to route based on a single identifier to maintain service transparency.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.4 R99 Terminals
|
See section 4.1.1 above: The following requirements for the support of R99 terminals is an operator specific option.
1. The standards shall enable the All-IP core network to support UMTS R99 terminals.
2. Speech services including emergency calls shall be provided in All-IP networks to any UMTS R99 terminal supporting these services.
3. To ensure roaming of non VoIP capable UMTS R99 terminals, speech services including emergency calls shall be possible based on CS capabilities of these terminals.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.5 Radio aspects
|
1. The radio resource usage should be optimised within the architecture for both service and signalling support.
2. Separation of the radio related and radio un-related functionalties between the core network and the radio access network
3. Separation of the user plane and control plane protocol stacks in the radio access networks
4. Fast uplink access and fast resource assignment procedures in both uplink and downlink for multiplexing different types of traffic on the same air link.
5. Optimization of end-to-end IP transport for certain class of real time applications (e.g. header compression, header stripping)
6. Network controlled handover procedure with short interruption to support real time applications (see handover requirement, section 4.9.)
7. Protocol stacks in the access network to support a range of services with different QoS requirements
8. Inter-working/interoperability of the QoS mechanism developed for the radio access network and the QoS mechanism used in the packet core network.
9. Bearer differentiation capability at the access network for multiplexing different types of traffic on the air to achieve maximum spectrum utilization
10. Optimal coding and interleaving for some applications such as voice.
11. Support of multiple data flows with different QoS per IP address (as defined in QoS framework in release 99)
12. Spectrum efficiency shall be maximized (e.g. statistical multiplexing).
13. The ERAN shall support GPRS and EGPRS services for pre-Release 2000 terminals.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.6 Interworking
|
1. The All-IP core network shall support interworking to external IP and non-IP networks (e.g. circuit-switched networks (PSTN, ISDN, GSM PLMN, UMTS PLMN,...).
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.7 Mobility management
|
1. The All-IP core network shall provide streamlining and CN operated hand-over procedures for UMTS.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.8 Roaming
|
1. The standard shall enable the All-IP core network to support roaming with 2G GSM/GPRS networks and R99 UMTS networks.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.9 Handover
|
The support of handover between release 98, release 99 and release 2000 network technologies is essential in maintaining adequate network coverage. Table 4-1 illustrates the necessary handover scenarios and the status of development of mechanisms.
Table 4-1: Handover requirements for UMTS All IP network
Between
2G-GSM cs
2G-GPRS
UMTS cs (R99)
UMTS ps (R99)
UMTS IP (PS services)
UMTS IP (CS services)
UMTS IP (PS services)
Req R00*
Req R00
Req R00*
Req R00
OK
Not Required
UMTS IP (CS services)
Req R00
Not Req
Req R00
Not Req
Not Req
OK
Key:
OK Same technology
Req R00 Required for release 2000
* The implications of the requirement for PS to/from CS handover in R00 are the subject of much debate. Alternatives for either the support of handover or to provide service coverage need to be investigated.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.9.1 Handover Categories
|
1 Intra network handover
Handover inside one all IP network
1a Intra RAN handover
1b Inter RAN handover
2 Inter network handover
Handover between two all IP networks
3 Inter-system handover
Handover between an all IP network and other systems
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.9.2 General Definition
|
Reselection and handover are two methods of supporting mobility during an active session. Reselection is the process whereby the mobile station autonomously determines which cell the mobile will receive services on. Handover is the process whereby the network determines which cell the mobile will receive services on.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.9.3 General Requirements
|
For real-time services handover procedure shall be used. The network shall control the handover procedure.
Handover shall be the selected method of mobility if one or more active sessions have requested handover in a multi session call with different QoS requirements
Performance requirement on speech interruption (i.e. “mute” period) shall be equivalent to or better than than GSM or ANSI-136 handover for telephony services. TBD for other services offered by an all IP Network.
Maintain maximum packet loss limit (i.e. less than TBD) and maximum delay limit (i.e., less than TBD) during handover.
Non-real-time services shall use either handover or cell reselection depending on the QoS parameters in combination with network parameters.
The Subscribed QoS level should be maintained across the handover boundary. However QoS negotiation (if neceassary) should be possible before, during and after the handover (The application may reject the offered QoS)
Handover procedure shall utilize radio resources efficiently.
Handover shall not compromise the security of: the network providing the new radio resources; the (possibly different) network providing the original radio resources; and the terminal UE.
There shall be efficient handling of multiple bearers, e.g. if voice and email transfer is going on simultaneously.
Essential IP/UDP/RTP header information (for inter and intra all IP network handover), as seen by an IP end point, shall be preserved across handover boundary. The required essential header information depends on the bearer.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.9.4 MS Requirements
|
The mobile station shall be capable of supporting both reselection and handover.
The mobile station shall aid the RAN in the handover decision by supplying RF environmental information (e.g. received signal strength from serving cell and neighbour cells).
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.9.5 RAN Requirements
|
Handover decisions shall be based in the RAN.
Maintain the RAN QoS parameters, associated with the mobile station, across a handover boundary. Note, RAN QoS parameters for a mobile station are based upon the negotiated set of QoS parameters.
Facilitate admission control to optimize radio resources.
Select a handover target based on criteria such as RF environmental information, radio resources of the neighboring cells, QoS requirements of the session, etc.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.10 Call Control and Roaming
|
The following requirements need consideration for call control and roaming support in an all IP based network.
1 Routing of signalling and transport needs to be optimised, for the purposes of call control and roaming between networks.
2 Whenever possible, tromboning of the user's voice or data communication session back to their home environment should not be used to provide the user with services when roaming outside their home network.
3 The Release 2000 all IP network must comply with the mandated requirements for Emergency Services.
4 The Release 2000 all IP network must comply with the mandated requirements for Number Portability.
5 The Release 2000 all IP network must support multiparty voice and data communications sessions including the capability for the user or service logic to dynamically add or delete users from an active communications session.
6 The Release 2000 all IP network must be able to accept and re-route incoming voice or data communication requests that are addressed to the user's directory number during periods of realignment of the national numbering plans (e.g., NPA splits in North America).
7 Transcoding of the traffic (voice, data, video) should be minimised. For example,, if the terminal equipment of the called and calling party have the same vocoder, no transcoding of the voice traffic, within the network, would occur.
8 The Release 2000 all IP network must provide connection to the services of the legacy 2G and release 99 networks.
9 The Release 2000 all IP network shall support VHE for roamers.
10 A minimum set of services for roamers shall be provided within the serving network. This minimal set of user services is still being defined. However, the following is anticipated to be in this minimal set of user services:
a Speech call and data session origination.
b Speech call and data session termination.
c Call Waiting for voice calls in the case of monomedia session
d Call Forwarding services for voice calls.
e Calling party identification information
f SMS
11 In the event that the Release 2000 all IP operator does not have a legacy network in the market that a Release 2000 all IP network is being deployed into and the Release 2000 all IP operator does not have any business relationships with the operators of the legacy networks. Consequently, the design of the Release 2000 all IP network can not assume that the requirements for mandated services or operator-specific services can be satisfied by forwarding the Release 2000 all IP call to the legacy network. The following are examples of Operator Services that may need to be handled directly by the Release 2000 all IP networks:
a Directory Assistance
b Third party billing
c Collect calls
d Calling card calls
12 When a Release 2000 all IP user roams from a Release 2000 all IP network to another Release 2000 all IP network and gets access to both transport services (e.g. GPRS) and application level services (e.g. multimedia calls), services may be provided by a CSCF in the serving network or by a CSCF in the home network. The serving network shall contain the information to contact the user's home network for the user's profile information. The CSCF of the serving network shall have access to the necessary information for the invocation and control of the user's advanced/ supplementary services at the user's home network.
13 The user shall be able to gain access to their ISPs or corporate LAN application level services..
14 Both dynamic and dedicated IP addresses shall be supported.
15 Release 2000 all IP networks will be capable of providing VPN functionality. VPN refers to both the GSM VPN and Intranets. The VPN features supported require further study and analysis.
16 When the UE uses the service of a CSCF/MSC server, the CSCF/ MSC Server needs to authenticate the UE. The way this is handled when the UE uses the services of a CSCF in the visited network requires further study and analysis.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
4.11 Security
|
The general principle for security for the all IP network implementation is to reuse the same mechanisms developed for 3GPP Release 99 wherever possible.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5 Architecture for an all IP PLMN
| |
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.1 Reference Architecture
|
The reference architecture provides two options:
Option 1: has been developed with the goal of allowing operators to deploy an all IP based architecture to deliver 3rd Generation wireless/mobile services. This architecture is based on packet technologies and IP telephony for simultaneous real time and non real time services.
Option 2:. One purpose of option 2 is also to allow support of release 99 CS domain terminals. In addition option 2 also supports the IP based services of option 1.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.1.1 Reference Architecture – Option 1
|
As described earlier in the Requirements section 4.1, the architecture shown in Figure 5-1 has been developed with the goal of allowing operators to deploy an all IP based architecture to deliver 3rd Generation wireless/mobile services. This architecture is based on packet technologies and IP telephony for simultaneous real time and non real time services.
The architecture shown and the components of which are described in subsequent sections allow for flexible and scaleable mechanisms to support global roaming and interoperability with external networks such as PLMN, 2G Legacy networks, PDNs and other multimedia VOIP networks.
The end-to-end architecture consists of the following key segments:
a) Radio Network
b) The GPRS network
c) The Call Control
d) Gateways to external networks
e) The Service architecture
The Radio network part consists of the equipment associated with the mobile user, the Radio Airlink and the Radio Access Network. The RAN supports both the UTRAN and the EDGE technologies.
Section 4 indicates that the intent of the core network part of the all IP architecture is that it should be designed to allow operators to use other access networks, for example ERAN and HIPERLAN 2. Within Figure 5-1, the ERAN is shown explicitly, where as the other access networks are represented by the bubble labelled “Alternative Access Network”. For the purposes of this report,
the ERAN is defined as an evolved GSM BSS supporting EDGE modulation schemes on a 200kHz basis and real time packet services
The support of alternative access networks, and the impact of the All IP architecture on the ERAN are outside the scope of this activity. However, to avoid the loss of information, the report does indicated where requirements are known to apply to these access networks.
Within this report, the reference point between the ERAN and the core network is designated as the Iu_ps’. That is, the reference point is Iu and the implementation is expected to be similar to that of the Iu_ps.
Note: the use of the reference label Iu_ps’ is confusing. During the standardisation activity a more suitable label should be chosen.
The GPRS network part has the GSNs which provide the mobility management and the PDP context activation services to the mobile terminal as they do in the R99 GPRS PS domain network. The HLR functionality for the GPRS network is provided by the Home Subscriber Server, (HSS).
The Call Control part of the architecture is the most critical functionality. The CSCF, MGCF, R-SGW, MGW, T-SGW and the MRF comprise the Call Control and signalling functionality to deliver the real-time mobile/wireless services. The CSCF is similar to the H.323 GateKeeper or a SIP Server. The architecture has been intentionally kept generic and is not based on a specific call control mechanisms such as H.323 or SIP. Such a choice is for further study.
The user profiles are maintained in the HSS. The signalling to the multimedia IP network is interface solely via the CSCF while the bearer is interfaced directly with the GGSN. The MRF interfaces with all bearer components for bearer media and with the CSCF for signaling. The MRF provides for media mixing, multiplexing, other processing and generation functions.
The interconnectivity to external networks such as PLMN, other PDNs, other multimedia VOIP networks and 2G Legacy networks (GSM or TDMA) is supported by the GGSN, MGCF, MGW, R-SGW and T-SGW functional elements. Other PLMNs are interfaced for both bearer and signalling via their respective GPRS components. The CSCF is a new component which also participates in this signalling. The signalling to legacy mobile networks is interfaced via the R-SGW, CSCF, MGCF, T-SGW and HSS, while the bearer is interfaced to and from the legacy PSTN network via an MGW. Legacy landline circuit switch signalling is interfaced via the CSCF, MGCF and T-SGW while the bearer is interfaced to and from the legacy PSTN network via an MGW.
The Service Architecture part of the network is currently depicted as an external entity and is described in detail in the section 10. Non-standard services are provided via interfaces to an application services layer. The HSS, the SGSN and the CSCF interface with the application and services bubble.
The details for each of the functional elements of the architecture are provided in subsequent subsections of this document.
Figure 5-1: Reference Architecture for Option 1
The Gm interface between the UE and the CSCF consists of the user to network multimedia signalling. It is carried on radio, Iu, Gn and Gi interfaces.
The SGSN and GGSN are the same functional elements as defined in 23.002 [4] for R99 of UMTS.
Mobility management procedures for Release 2000 all IP network MSs in Release 2000 all IP networks are based on the Routing Area identifier (RAId). Mobility management procedures for CS capable MSs in Release 2000 all IP networks are based on the Routing Area identifier (RAId) and on Location Area Identifiers. In case of roaming between Release 2000 all IP networks to 2G and vice versa, a mechanism to convert the format of identities is needed (i.e. MS roaming from Release 2000 all IP network to 2G network has knowledge only of old RA but Location Area Identifier also needs to be given to the 2G-MSC). When roaming from 3G-R00 to 2G (without possibility of combined update), how the 2G-MSC can retrieve the IMSI from the all IP core network is an open issue.
MS identity (IMSI) is protected over the radio interface through the adoption of a single temporary identifier (P-TMSI) allocated by the SGSN during the MS registration procedure. P-TMSI can be reallocated at every following registration or routing area update. [relevance to roaming scenarios between Release 2000 all IP networks and legacy cellular networks]
The Access Network Nodes (GSN(s), RNC) are not aware the multimedia signaling protocol between the UE and the CSCF. They are even not aware that a given UE sends or does not send signalling to the CSCF.
Note1: this does not preclude that to optimize the radio, the RNC might support specific RAB for the individual flows of the multimedia user plane. These RAB are requested by the UE at PDP context activation.
Note2: the SGSN may have a role in the choice of the CSCF by the UE. This is FFS. But even if the SGSN participate in the CSCF address determination, the SGSN does not carry out the multimedia registration on behalf of the UE.
Different PDP contexts carry multimedia signalling and user flows due to different requirements on QoS for these PDP contexts. The Access Network Nodes (GSN(s), RNC) are nevertheless not aware whether a given PDP context carries multimedia signalling or not.
Working Architectural approach
1. The All IP Core network is engineered primarily to use a common technology (IP) to support all services including multimedia and voice services controlled by H.323/SIP or ISUP.
2. Network architecture is based upon IP packet technologies for simultaneous real-time and non-real-time services.
3. Network architecture is based upon an evolution of GPRS.
4. For support of R99 CS domain services the R99 CS domain CC mechanism may be reused. (NOTE: This does not prevent alternative mechanism such as H.323, SIP or evolved forms of R99 CS domain CC mechanisms being used by operators to deliver R99 CS domain services)
5. For the support of release 00 terminals are IP based, and the integration of services is obtained through IP.
6. Network architecture should support personal mobility and interoperability between mobile and fixed networks for both voice and data services.
7. Maintain or improve quality of service levels when compared to today’s networks.
8. Maintain or improve network reliability when compared to today’s networks.
9. All IP interfaces and associated network interfaces should be enhanced to support real-time multimedia services.
10. Network architecture will provide a separation of service control from call/connection control.
11. Network architecture will replace SS7 transport with IP.
12. Network architecture will be independent of network transport layers of Layer 1 (L1) and Layer 2 (L2).
13. Regardless of service type, ISUP based or IP based, IP transport shall be possible for all signalling and data transport.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.1.2 Reference Architecture – Option 2
|
As described earlier in the Requirements section 4.1.1 item 3 and 6, the architecture shown in Figure 5-2 allows operators to migrate from a R’99 UMTS network into a R’00 All IP network. One purpose of option 2 is to allow support of release 99 CS terminals. Option 2 allows the two domains of R’99 to evolve independently.
As for option 1 the architecture of option 2 allows that all services supported by option 2 share bearer level transport and bearer control. Various underlaying transport mechanisms shall be allowed (e.g. RTP/UDP/IP, AAL2/ATM or STM).
Reference architecture option 2 includes the SGSN/GGSN/CSCF based services of reference architecture option 1. The definitions and working architectural approach described in chapter 5.1.1 therefore also applies to the SGSN/GGSN/CSCF part of option 2.
[Note: The requirement in section 4.3 for the support of routing on a single MSISDN or to allow operators to move subscribers from the CS services to the All IP service without changing MSISDN will be solved as part of R99.]
Two control elements, related with R’99 CS domain architecture, are added in option 2; the MSC server and GMSC server.
Option 2 benefits from the Iu architecture of R’99, having transport of user data separated from control, to allow UTRAN to access the core network via a MGW separated from the MSC server. Between UTRAN and MSC server the control part of Iu, RANAP, is used.
[Note: 1) Exact definition of ERAN in relation to GSM/teleservice speech and ERAN’s relations to MSC server/MGW needs to be defined. 2) The possibility of interfacing GSM/BSS to MGW and MSC server shall be further studied.].
By allowing servers to terminate MAP and the user-network signalling (04.08+ CC+MM), requirements related to service and network migration of R’99 UMTS CS domain services and network migration can be fulfilled. The requirements that requires a network architecture according to option 2 are:
Section Requirement Number
4.1.1 2 (to meet the timescales of R00)
4.1.1 3
4.1.1 6
4.4 all
4.8 1 (the need for option 2 to meet this requirement will depend upon the roaming solution chosen)
Figure 5-2: Reference Architecture for Option 2
Iu is the reference point between UTRAN and all IP core network. Between UTRAN and SGSN Iu is IP based. Between UTRAN and MGW - Iucs (RTP, AAL2) - Iu may be based on different transport technologies.
MAP is operated between HSS and MSC server and GMSC server respectively.
5.1.3 What is the Border of the Network
Open Issue: What is the Border of the Network
In the case that the GGSN is seen as the border of the network towards the IP network or the GGSN+MGW is seen as the network border towards the PSTN/Legacy network, then the following issue need to be clarified: how to determine the MGW and how to ensure the most optimal routing towards this MGW. On call setup, the CSCF needs to determine the appropriate MGW for the call. For example, it needs to determine if the call is to the PSTN (and in this case towards which PSTN network) or to a Voice Over IP network. This determination can only be ensured when call set-up signalling has been analysed by the CSCF and possibly by the SCF. This analysis may change the called party number (for instance modify the called party address from the address corresponding to an IP terminal to an address corresponding to a foreign PSTN terminal). It is only at that point that the best MGW can be determined. This determination cannot be done before by the SGSN. The MGW address is sent back (through H.225 signalling) to the UE. The UE can then activate a PDP context (for the support of user plane traffic) towards the appropriate network (i.e. the network that best allows to reach the MGW to be used by the call).
Note [FFS]. The issue of optimal routing when an MRF is added cannot be determined until there is agreement on what is the border of the network.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.2 New Functional Elements
|
5.2.1 Call State Control Function (CSCF)
In the following section, CSCF has been divided into several logical components.
Currently, these logical components are internal to the CSCF. The need for external components to be able to address directly one of the logical components of the CSCF is for FFS.
Every CSCF acting as a Serving CSCF (see section 9) has a CCF function.
ICGW (Incoming call gateway)
• Acts as a first entry point and performs routing of incoming calls,
• Incoming call service triggering(e.g. call screening/call forwarding unconditional) may need to reside for optimisation purposes,
• Query Address Handling (implies administrative dependency with other entities)
• Communicates with HSS
CCF (Call Control Function)
• Call set-up/termination and state/event management
• Interact with MRF in order to support multi-party and other services
• Reports call events for billing, auditing, intercept or other purpose
• Receives and process application level registration
• Query Address Handling (implies administrative dependency)
• May provide service trigger mechanisms (service capabilities features) towards Application & services network (VHE/OSA)
• May invoke location based services relevant to the serving network
• May check whether the requested outgoing communication is allowed given the current subscription.
SPD (Serving Profile Database)
• Interacts with HSS in the home domain to receive profile information for the R00 all-IP network user and may store them depending on the SLA with the home domain
• Notifies the home domain of initial user’s access (includes e.g. CSCF signalling transport address, user ID etc. needs further study)
• May cache access related information (e.g. terminal IP address(es) where the user may be reached etc.)
AH (Address Handling)
• analysis, translation, modification if required, address portability, mapping of alias addresses
• May do temporary address handling for inter-network routing.
Other functions such as admission control, multiple session knowledge within one CSCF, multiple CSCFs serving one terminal for multiple services, role of multiple CSCFs serving a network etc. need further investigation.
Interfaces need to be further studied and defined.
5.2.2 Home Subscriber Server (HSS)
The Home Subscriber Server (HSS) is the master database for a given user. It is responsible for keeping a master list of features and services (either directly or via servers) associated with a user, and for tracking of location of and means of access for its users. It provides user profile information, either directly or via servers. It is a superset of the Home Location Register (HLR) functionality., for example as defined in GSM MAP, but differs in that it needs to also communicate via new IP based interfaces. The HSS shall support a subscription profile which identifies for a given user for example:
• user identities
• subscribed services and profiles
• service specific information
• mobility management information
• authorization information
Like the HLR, the HSS contains or has access to the authentication centers/servers (e.g. AUC, AAA).
Figure 5-3: Example of a Generic HSS structure and basic interfaces
Figure 5-4: Example of HSS structure with UMS Specific Functionality
The HSS may consist of the following elements as shown in the Figure 3:
1) User Mobility Server (UMS): it stores the Release 2000 all IP network Service Profile(see section 9.1) and stores Service Mobility or Serving CSCF related information for the users. UMS might also generate, store and/or manage security data and policies (e.g. IETF features). UMS should provide logical name to transport address translation in order to provide answer to DNS queries. UMS role and functional decomposition are for further study.
2) 3G HLR: A GPRS HLR enhanced to support Release 2000 all IP networks GPRS specific information.
Gr and Gc use MAP which may be implemented using MAP transported over IP, however the issue of roaming to a network that supports MAP over SS7 needs to be considered. The Cx interface requires further study: it may be implemented via IETF protocols such as DNS or via MAP procedures.
Following functionality may need to be supported in the HSS and are for further study:
• it stores the R00 all- IP network Service Profile and stores location information for the users.
• it may also generate, store and/or manage security data and policies (e.g. IETF features).
• may need to provide logical name to transport address translation.
• The HSS interacts with the R-SGW to communicate with VLRs and other Mobility managers which do not use IP. HSS interfaces with CSCF via Cx which is for further study.
• Other R00 all-IP based network functions such as AAA, DNS etc. and their interactions with HSS is for further study.
• Interface(s) between UMS and 3G HLR is for further study.
note: If the user profile is split across different databases then there should either be no duplication of information elements or the consistency of the data should be maintained.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.2.3 Transport Signalling Gateway Function (T-SWG)
|
This is component in the R00 all-IP network is PSTN/PLMN termination point for a defined network. The functionality defined within T-SGW should be consistent with existing/ongoing industry protocols/interfaces that will satisfy the requirements.
• Maps call related signalling from/to PSTN/PLMN on an IP bearer and sends it to/from the MGCF.
• Needs to provide PSTN/PLMN <-> IP transport level address mapping.
Interfaces need to be further studied and defined.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.2.4 Roaming Signalling Gateway Function (R-SGW)
|
The role of the R-SGW described in the following bullets is related only to roaming to/from 2G/R99 CS and GPRS domain to/from R00 CS and GPRS domain and is not involving the multimedia/VoIP domain.
• In order to ensure proper roaming, the R-SGW performs the signaling conversion at transport level (conversion: Sigtran SCTP/IP versus SS7 MTP) between the legacy SS7 based transport of signaling and the IP based transport of signaling. The R-SGW does not interpret the MAP / CAP messages but may have to interpret the underlying SCCP layer to ensure proper routing of the signaling.
• (For the support of 2G / R99 CS terminals): The services of the R_SGW are used to ensure transport interworking between the SS7 and the IP transport of MAP_E and MAP_G signalling interfaces with a 2G / R99 MSC/VLR
For the Multimedia/VoIP domain, MAP interworking at the R-SGW is for Further Study.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.2.5 Composite Gateway
|
Composite gateway: A logical entity composed of a single MGC and one or more MGs that may be reside on different machines. Together, they preserve the behaviour of a gateway as defined in H.323 and H.246 (this may include SIP servers and MSC servers in release 2000).
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.2.6 Media Gateway Control Function (MGCF)
|
This component in the R00 all-IP network is PSTN/PLMN termination point for a defined network. The functionality defined within MGCF should be consistent with existing/ongoing industry protocols/interfaces that will satisfy the requirements.
• Controls the parts of the call state that pertain to connection control for media channels in a MGW.
• Communicates with CSCF.
• MGCF selects the CSCF depeneding on the routing number for incoming calls from legacy networks.
• Performs protocol conversion between the Legacy (e.g. ISUP, R1/R2 etc.) and the R00 all-IP network call control protocols (this is still under further study within the industry).
• Out of band information assumed to be received in MGCF and may be forwarded to CSCF/MGW.
Interfaces need to be further studied and defined.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.2.7 Media Gateway Function (MGW)
|
This component in the R00 all-IP network is PSTN/PLMN transport termination point for a defined network. For the architecture option 2, the component is also used for interfacing UTRAN with the All IP core network over Iu.
The functionality defined within MGW should be consistent with existing/ongoing industry protocols/interfaces that will satisfy the requirements.
A MGW may terminate bearer channels from a switched circuit network (i.e., DSOs) and media streams from a packet network (e.g., RTP streams in an IP network). Over Iu MGW may support media conversion, bearer control and payload processing (e.g. codec, echo canceller, conference bridge) for support of different Iu options for CS services: AAL2/ATM based as well as RTP/UDP/IP based. [Note: in the general R’00 architecture different core network transport technologies shall be possible for example: ATM, STM or IP.]
• Interacts with MGCF, MSC server and GMSC server for resource control.
• Owns and handles resources such as echo cancellers etc.
• May need to have codecs.
In band signalling impacts to MGW and R00 all-IP network is for further study.
Functionality on the delivery of ring tone towards PSTN/ PLMN are for further study.
The MGW will be provisioned with the necessary resources for supporting UMTS/GSM transport media. Further tailoring (i.e packages) of the H.248 may be required to support additional codecs and framing protocols, etc.
For architecture option 2, the MGW bearer control and payload processing capabilities will also need to support mobile specific functions such as SRNS relocation/handover and anchoring (note that these functions are provided by the SGSN/GGSN in architecture option1 and are not required in the MGW). It is expected that current H.248 standard mechanisms can be applied to enable this. Solutions of how to use the H.248 generic bearer control mechanisms for mobile specific functions needs further studies.
Interfaces need to be further studied and defined.
5.2.8 Multimedia Resource Function (MRF)
• This component performs multiparty call and multi media conferencing functions. MRF would have the same functions of an MCU in an H.323 network.
• Responsible for bearer control (with 3G-GGSN and MGW) in case of multi party/multi media conference
• May communicate with CSCF for service validation for multiparty/multimedia sessions.
Handling of resources such as two stage dialling, announcements etc. are for further study.
Interfaces need to be further studied and defined.
5.2.9 MSC Server
MSC server mainly comprises the call control and mobility control parts of a GSM/UMTS R99 MSC.
The MSC Server is responsible for the control of mobile originated and mobile terminated 04.08CC CS Domain calls. It terminates the user-network signalling (04.08+ CC+MM) and translates it into the relevant network – network signalling. The MSC Server also contains a VLR to hold the mobile subscriber's service data and CAMEL related data.
MSC server controls the parts of the call state that pertain to connection control for media channels in a MGW.
5.2.10 Gateway MSC Server
The GMSC server mainly comprises the call control and mobility control parts of a GSM/UMTS R99 GMSC.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.3 Description of Reference Points
| |
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.3.1 Cx Reference Point (HSS – CSCF)
|
This reference point supports the transfer of data between the HSS and the CSCF.
When a UE has registered with a CSCF, the CSCF can update its location towards HSS. This will allow the HSS to determine which CSCF to direct incoming calls to. On this update towards the HSS, the HSS sends the subscriber data (application related) to CSCF.
For a MT call, CSCF asks the HSS for call routing information.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.3.2 Gm Reference Point (CSCF – UE)
|
This interface is to allow UE to communicate with the CSCF e.g.
• register with a CSCF,
• Call origination and termination
• Supplementary services control.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.3.43 Mc reference point (MGCF – MGW)
|
The Mc reference point describes the interfaces between the MGCF and MGW, between the MSC Server and MGW, and between the GMSC Server and MGW. It has the following properties:
• full compliance with the H.248 standard, baseline work of which is currently carried out in ITU-T Study Group 16, in conjunction with IETF MEGACO WG.
• flexible connection handling which allows support of different call models and different media processing purposes not restricted to H.323 usage.
• open architecture where extensions/Packages definition work on the interface may be carried out.
• dynamic sharing of MGW physical node resources. A physical MGW can be partitioned into logically separate virtual MGWs/domains consisting of a set of statically allocated Terminations.
• dynamic sharing of transmission resources between the domains as the MGW controls bearers and manage resources according to the H.248 protocols.
For architecture option 2, the functionality across the Mc reference point will need to support mobile specific functions such as SRNS relocation/handover and anchoring. It is expected that current H.248/IETF Megaco standard mechanisms can be applied to enable this. Solutions of how to use the H.248 generic bearer control mechanisms for mobile specific functions needs further studies.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.3.4 Mh Reference Point (HSS – R-SGW)
|
This interface supports the exchange of mobility management and subscription data information between HSS and R99/legacy mobile networks. This is required to support All IP users who are roaming in a 2G network.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.3.5 Mm reference Point (CSCF – Multimedia IP networks)
|
This is an IP interface between CSCF and IP networks. This interface is used, for example, to receive a call request from another VoIP call control server or terminal.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.3.6 Mr Reference Point (CSCF - MRF)
|
Allows the CSCF to control the resources within the MRF
5.3.7 Ms reference Point (CSCF – R-SGW)
This interface allows CSCF to contact legacy network elements, e.g. 2G HLR, for location management (location update and subscriber data download), and call control (eg 2G HLR enquires for routing number (RN) for a roaming 2G user).
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.3.8 Mw Reference Point (CSCF – CSCF)
|
The interface allows one CSCF (e.g. home CSCF) to relay the call request to another CSCF (eg serving CSCF).
5.3.9 Nc Reference Points (MSC Server – GMSC Server)
Over the Nc reference point the Network-Network based call control is performed. Examples of this are ISUP or an evolvement of ISUP for bearer independent call control (BICC). In the all IP core network Nc reference point uses an IP based signalling transport. [Note: in the general R’00 architecture different options for signalling transport on Nc shall be possible.]
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
5.3.10 Nb Reference points (MGW-MGW)
|
Over the Nb reference point the bearer control and transport are performed. The transport may be RTP/UDP/IP or AAL2 for transport of user data. The bearer control over Nb is FFS, it may be based on RTP, H.245 or corresponding. [Note: in the general R’00 architecture different options for user data transport and bearer control shall be possible on Nb, for example: AAL2/Q.AAL2, STM/none, RTP/H.245.]
5.3.11 SGSN to Applications and Services
The interface from the SGSN to the SCP in the Applications and services domain is the interface defined for GPRS to support Charging Application Interworking.
5.4 Usage of MAP/CAP - Protocol stack below MAP / CAP – General considerations
Below MAP and CAP, the protocol stack within the All IP CN is as shown in Figure 5-5:
• SCCP and TCAP are used below CAP and MAP. Indeed CAP and MAP both rely on services provided by these underlying protocols (e.g., transaction capabilities, global title translation). Alternatives to providing the services of SCCP is for further study.
• The lower transmission layers are compliant with the IETF Sigtran protocol suite used to carry telecommunication signalling on top of an IP backbone.
Figure 5-5: All IP R00 protocol stack for MAP/CAP
This protocol stack is used to carry MAP/CAP flows:
• inside the All IP CN, between nodes terminating the MAP/CAP: e.g. between HSS or SCP and the All IP CN functions (SGSN, MSC servers …) handling Call / Session and needing to dialog with the HSS or SCP for user mobility management / subscriber data retrieval. This is for example the case of the Gr, Gc interfaces.
It may also be envisaged (although this requires further study) to use MAP on Cx for user mobility management / subscriber data retrieval.
• in case of interworking with nodes not supporting this MAP/CAP over IP stack (e.g. 2G or UMTS R99 networks nodes) but needing a MAP / CAP dialog with a node supporting this MAP/CAP over IP stack. This protocol stack is used between the nodes terminating MAP/CAP and the R-SGW. The Mh interface shall use MAP over IP, for the roaming scenarios involving the R00 CS domain and GPRS (excluding the VoIP/Multimedia domain), hence this does not exclude other protocols being used. The use of MAP on the Ms is FFS.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
6 QoS
|
The work currently being done within the S2 QoS Ad Hoc is reflected within TR 23.907 and the QoS section of TR 22.105. The R99 version of these specifications will support real time applications on a packet switched network which includes the ability of UMTS to transparently support multi‑media applications that utilize the H.323 protocol. The all‑IP architecture as described in this document defines the implementation of a call control function which can be based on either SIP or H.323 within the PLMN. Therefore, the R00 QoS work will include any changes required to support QoS capabilities necessary for support of multi‑media based on H.323 or SIP within the PLMN. Doing this is not expected to introduce any new QoS requirements at the UMTS bearer level.
In additon, please note that the desire also exists to have the all‑IP architecture support multi‑media applications that utilize the SIP protocol. However, QoS work related to the SIP protocol would only be undertaken within 3GPP when 3GPP itself undertakes the work to support the SIP protocol. However, we do anticipate that the QoS work centered on H.323 is directly applicable to SIP.
In addition, since the work on the all IP network includes EDGE support as identified by the ERAN architecture, the QoS work within the ERAN needs to be in alignment with QoS support within UMTS.
A preliminary review of the current version of TR 23.907 leads us to believe that it is largely sufficient to meet the objectives an all IP network. The review includes the following observations :
• TR 23.907 includes the specification of a QoS conversational class which includes voice. TR 23.907 identifies the fundamental characteristics of this class as:
• Preserve time relation (variation) between information entities of the stream.
• Conversational pattern (stringent and low delay)
• These characteristics apply whether voice is carried within a circuit or as packets. So there should be no need to modify the QoS classes currently defined.
• Implementing the H.323 call model within the PLMN is not expected to affect the R99 TR 23.907 identified QoS technical requirements, the overall architecture, nor the functions identified therein. However, a brief study will be necessary to verify this.
• The Radio Access Bearer Service attributes currently defined will need to be reviewed in light of an all IP network but minimal additions in this area would be expected for UMTS. However, for EDGE, work should be anticipated. Obviously the mapping from bearer to radio bearer is also affected.
• The issue of interworking the packet voice capable GPRS with other networks needs to be studied at least as it pertains to an acceptable voice delay budget.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
7 Handover
|
Within this section, the topics to be studied and standardised to support handover for real time services in the PS domain have been identified. This section has investigated various handover scenarios, however the fact that the scenario has been studied here does NOT imply a requirement for the support of that scenario. The requirements for handover relating to the All IP architecture of R00 in UMTS will be determined by S1 as part of the R00 Service Requirement specification work.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
7.1 SRNC Relocation within a UMTS R00 IP network
|
Within UMTS, work has already been undertaken to provide handover for real time PS domain services. The UTRAN does not distinguish between circuit and packet services, it simply provides for real and non-real time services, hence Intra RAN handover for real time services is available.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
7.1.1 Support Required within ERAN
|
The goal of the All IP architecture is to provide a common core network for both UTRAN and ERAN. The specification of this work is outside the scope of this study, however, it is worth noting that the ERAN will need to support the following procedures:
• SRNC Relocation
• Mobile Assisted Network Controlled handover for the real time packet services.
|
8a9283a656db20cd080be5c53231664c
|
23.922
|
7.1.2 All IP UTRAN to All IP ERAN Handover
|
The need to support this handover scenario is for FFS.
In this scenario, a CSCF will support terminals in both the ERAN and the UTRAN. The terminal will have access to the same Media Gateway from both RANs, hence the same media codec will be used in the network.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.