3G-324M Spec: Part 2

Understanding the 3G-324M Spec: Part 2

Eli Orr, Radvision
1?28, 2003 (3:32 $?)
URL: http://www.commsdesign.com/showArticle.jhtml?articleID=16500089


The age of an all-IP wireless network could be years away. However, the need for delivering multimedia services over wireless links is at an all-time high. To bridge this gap, the 3GPP and 3GPP2 standards bodies are converging on the 3G-324M specification for supporting video transmissions over wireless lines.


3G-324M supports the real-time streaming of multimedia broadband wireless communications by routing traffic over the circuit switched network, instead of the IP network. Being circuit-switched based, the standard has all the hallmarks of a protocol ideal for streaming real-time multimedia, including a fixed delay, low overhead of codecs, and no IP/UDP/RTP header overheads.


This is the second installment in our tutorial on the 3G-324M. In Part 1, we explored error resilience and concealment techniques, H.223 multiplexing/demultiplexing, and the 3G-324M adaptation layers. In Part 2, we'll examine H.245 support, voice coding/decoding, and the video channel. We'll also provide insight into some real-life implementation issues. Let's kick off our discussion with a look at the problem with delivering multimedia over 3G links.


H.245 Terminal Control Protocol
3G-324M uses H.245 as a terminal control protocol. H.245 is used by H.323 as well as by H.324 for PSTN and by H.310 for ATM. Currently (Q4/2002) ITU H.245 version 9 is ratified by the SG16 of ITU.


The oldest version of H.245 that can be supported in a 3G-324M implementation is version 3. However it is highly recommended to support higher version such as 6 or 7 which support by most implantations today or even upper for richer set of call control services of commands and indications. H.245 is fully backward compatible so that higher version can operates with lower version of H.245 and vise versa.


Since 3G-324M rides on a channel opened between two communicating parties it does not need any addressing such as in H.323. With this fact, it is expected that the gateway (e.g. between 3G-324M, H.320, H.323 and SIP) will provide the interoperability between different networks can be realized rather easily.


Since H.323 is not needed, H.245 requires the numbered simple retransmission protocol (NSRP) and control channel segmentation and reassembly layer (CCSRL) sublayer support to ensure reliable operation. H.245 requires mobile terminals to support NSRP and SRP modes. If both terminals start the session in level 0, H/245-enabled systems must operate in the SRP. If the terminals start a session at level 2, NSRP mode is employed. CCSRL, on the other hand, is used for carrying H.245 large packets required for operation.


In addition to providing NSRP and CCSRL support, the H.245 control protocol provides following functionalities and services:




Master-slave determination is provided to determine which terminal is the master at the beginning of the session. Due to the fact that H.245 is symmetric control protocol, it is necessary to decide the master terminal, which has the right to decide the conditions in case of the conflict.
Capability exchange is provided to exchange the capabilities both terminal supports, such as optional modes of multiplexing, type of audio/video codecs, data sharing mode and its related parameters, and/or other additional optional features.
Logical channel signaling is provided to open/close the logical channels for media transmission. This procedure also includes parameter exchange for the use of this logical channel.
Multiplex table initialization/modification is provided to add/delete the multiplex table entries.
Mode request is provided to request the mode of operation from the receiver side to the transmitter side. In H.245, the choice of codecs and its parameters are decided at the transmitter side considering decoder's capability, so if the receiver side has a preference within its capability, this procedure is used.
Round-trip delay measurement is provided to enable accurate quality characteristic measurement.
Loopback testing is provided for use during development or in the field to assure proper operation.
Miscellaneous call control commands and indications are provided to request the modes of communication, flow control such as conference commands, jitter indication and skew, or to indicate the conditions of the terminal, to the other side.

H.245 uses the abstract syntax notation 1 (ASN.1) to define each message parameters that provides readability and extensibility effectively. To encode these ASN.1 messages into binary, the packed encoding rule (PER) is used to realize the very bandwidth effective message transmission. As mentioned before after the multiplexing level synchronization between communicating parties is completed the first logical channel opened (channel 0) is H.245 call control with the CCRL and NSRP to assure that the H.245 channel will be highly reliable and can use large packets during operation.


Voice Channel—The AMR Codec
The 3G-324M specifications define the AMR codec as mandatory. 3G-324M also recommends the use of G.723.1, which is used by many H.323 terminals today.


The AMR codec was originally developed and standardized by the ETSI for GSM cellular systems. The AMR codec, rolling out in networks and terminals, dynamically adjusts the amount of bits allocated to voice coding and error control, providing the best possible voice quality at each instance based on radio conditions. AMR significantly enhances the effectiveness of frequency hopping and tighter reuse patterns by allowing a greater percentage of radio channels to be in use simultaneously, resulting in an additional capacity gain of about 150%.


AMR was chosen by 3GPP as the mandatory codec for 3G cellular systems. The AMR codec includes eight narrowband codec modes: 12.2, 10.2, 7.95, 7.4, 6.7, 5.9, 5.5 and 4.75 kbit/s. It also supports comfort noise (CN) for a discontinuous transmission (DTX) operational mode.


Besides the adaptation of rate, the AMR codec also supports unequal bit-error detection and protection (UED/UEP). The UEP/UED mechanisms allow more speech over a lossy network by sorting the bits into perceptually more and less sensitive classes. A frame is only declared damaged and not delivered if there are bit errors found in the most sensitive bits. On the other hand, acceptable speech quality results if the speech frame is delivered with bit errors in the less sensitive bits, based on human aural perception. An important characteristic for high BER environment such as wireless network is AMR's robustness for packet loss, through redundancy and bit errors, sensitivity sorting. Another benefit of AMR is the adaptive rate adaptation for switching smoothly between codec modes on the fly.


The Video Channel
The 3G-324M standard calls out the H.263 codec as mandatory and MPEG-4 as recommended codec for video processing. However, MPEG-4 is the 3G-324M standard de-facto used by all major supporting vendors. Resiliency and high efficiency make MPEG-4 codec particularly well suited for 3G-324M.


H.263 is a legacy codec that is used by many existing H.323 wire lined devices. MPEG-4 is much more flexible and offers advanced error detection and correction services, which are a big value add when delivering video over a wireless network. Let's look at the error detection and correction services in more detail.


When supported, 3G-324M says that MPEG-4 visual codecs shall support simple profile 1 level 0. MPEG-4 visual (ISO/IEC 14496-2) is a generic video codec. One of its target areas is mobile communications.


Error resiliency and high efficiency make the MPEG-4 visual codec particularly well suited for 3G-324M. MPEG-4 visual is organized into profiles. Within a profile, various levels are defined. Profiles define subsets of tool sets. Levels are related to computational complexity. Among these profiles, the simple visual profile provides error resilience (through data partitioning, RVLC, resynchronisation marker, and header extension code) and low complexity. MPEG-4 allows various input formats, including general formats such as QCIF and CIF. It is also baseline compatible with H.263.


As stated above, error resilience is achieved through resynchronization, byte alignment, data partitioning the reversible variable length code (RVLC), adaptive intra refresh (AIR), and error detection and concealment. Let's look at each of these in more detail.


1. Resyncrhonization: Under the MPEG-4 spec, a resynchronization marker can reduce the error propagation caused by the nature of variable length code (VLC) into single frame. In MPEG-4, the resynchronization marker is inserted at the top of a new group of blocks GOB with the header information (multiplexed block number [MBN], quantization parameters) and optional HEC, so that decoding can be done independently. It is a good idea to place the resynchronization marker prior to important objects like people to improve error resilience with minimum increase of overhead.


2. Byte alignment: Bit-stuffing for the byte alignment gives additional error detection capability through its violation check.


3. Data partitioning: A new synchronization code named motion marker separates the motion vector (MV) and discrete cosine transform (DCT) field to prevent from inter-field error propagation, thus allowing effective error concealment to be performed. When errors are detected solely in the DCT field, that multiplexed block (MB) will be reconstructed using correct MV. This results in natural motion better than simple MB replacement of the previous frame.


4. RVLC: The RVLC enables forward and backward decoding without significant impact on coding efficiency. This feature localizes error propagation ideally into single MB.


5. AIR: Different from the conventional cyclic intra refresh, AIR employs motion-weighted intra refresh, which results in better perceptual quality with quick recovery in corrupted objects.


6. Error detection and concealment: Errors can be detected through exception or violation in the decoding process, and then concealment will be applied. The functionality is included for mobile application. The endpoint of H.324 can support for MPEG-4 audio, so that MPEG-4 audio could be used for H.324 mobile phone terminal.


Integrating 3G-324M With Other Multimedia While 3G-324M is a straightforward protocol to implement in end devices and media servers, designers will face challenges making 3G-324M with other protocols such as H.323 and SIP. Let's look at this issue in more detail.


H.323 is based on Q.931 for call setup and H.245 for call control. 3GPP defines TS.26.112 for call setup procedure in UMTS. The interworking device shall map theTS-26.112 call setup into Q.931 H.323 calls and vise versa. For call control mapping, since both protocols uses H.245 the mapping is trivial, however the H.245 in 3G-324M is addressless. Thus a transcoding function may also be required to ensure that 3G-324M works with various H.323 devices supporting codecs such as H.261 and H.263.


SIP is based of session description protocol (SDP) for both call setup and call control. Hence both TS 26.112 and 3G-324M H.245 call control should be mapped into SDP messages and vise versa. Again a transcoding functions may be required to ensure that 3G-324M systems work with SIP-based systems.


Editor's Note: To view Part 1 of this article,