Mobile Networks for Video Telephony

Mobile Networks for Video Telephony

In this section we review briefly the mobile network architectures and options that enable mobile video telephony. The considerations made in Chapter 4, Section 3 remain valid also for the case of mobile video telephony. However, because this type of application is more challenging in terms of end-to-end delays, not all the network configurations presented in Chapter 4 are suited for mobile video telephony. Therefore the main purpose of this section is to select the mobile network channels that enable video telephony.

Mobile channels can be divided into two categories:

1. Circuit-switched (CS) channels

2. Packet-switched (PS) channels

Table 21.1 shows a summary of network channels that can enable mobile video telephony (the bit rates indicated are maximum, and practical mobile videophone terminal implementations can have even lower maximum bit rates). In this table, we find neither GPRS Release '97 nor EGPRS networks. The reason is that they are not capable of sustaining conversational real-time traffic because of the high delay bounds compared to those required to support video telephony services.

Table 21.1: Mobile Network Channels for Video Telephony

Mobile Network	CS/PS	Theoretical Maximum Bit Rates (kbps)	Layer 2 Configuration
HSCSD	CS	57.6	Transparent mode
ECSD	CS	64.0	Transparent mode
UMTS (UTRAN) Release 99 and Release 4	CS	64.0	Transparent mode
UMTS (UTRAN) Release 5	PS	2048.0	Unacknowledged mode
UMTS (GERAN) Release 5 Gb mode	PS	473.6	Unacknowledged mode

For implementing mobile video telephony both CS and PS bearers can be used. In either case the transmission channel must be transparent. This implies that no retransmissions or mechanisms that produce additional delays must be employed at layer 2 of the mobile network (data link layer). In fact, layer 2 protocol data units (PDUs) are required to have the smallest header overhead, in order to reduce (or totally avoid) processing delays induced by complex PDU encapsulation and decapsulation. Unacknowledged mode is generally used at the data link layer in packet-switched connections. The PDUs used are slightly more complex than those used for the transparent mode, but light enough to allow fast data delivery between the two layer-2 peer entities.

UMTS networks allow theoretical maximum bit rates of 2048 kbps. However, the tested CS connections for video telephony for Release '99 and Release 4 networks are up to 64 kbps, as defined in 3GPP. ^[2], ^[3] In these specifications, the recommended bit rates for video telephony services are 32 and 64 kbps, whereas the offered residual BERs are in the order of 10–4 or 10–6. For PS connections, the maximum bit rates indicated in Table 21.1 are just theoretical. In practice, the tested and implemented maximum bit rates will be much smaller (in the order of 384 kbps).

The QoS profile for conversational traffic is defined in the 3GPP specification. ^[4] It is very similar to the profile defined for streaming traffic in Chapter 4, Section 3.1.2. However, due to the more-stringent delay requirements for the conversational traffic, two key parameters need to be defined differently:

1. Service data unit (SDU) error ratio. The maximum value for this parameter is defined as 10–2. In other words, whenever erroneous packets are not delivered to the higher protocol layers and are considered lost the maximum packet loss rate is equal to 1 percent. The corresponding value defined in the QoS profile for streaming traffic is ten times larger, i.e., 10 percent. The rationale behind this parameter selection is that a higher packet loss rate can be allowed for streaming traffic. However, making use of higher-layer retransmissions, which can be implemented because streaming traffic can tolerate larger end-to-end delays, can reduce this error rate. Whenever no retransmissions are allowed, such as in the case of conversational traffic (i.e., video telephony traffic), a smaller maximum SDU error rate is more appropriate, and conservatively it helps in yielding a better application QoS.

2. Transfer delay. Because the end-to-end delay requirements for mobile video telephony are more stringent than for streaming service, the QoS profile defined for conversational traffic includes delay values that are more challenging than those allowed for streaming (where lower bounds are equal to 280 ms). For conversational traffic, the lower bound for UMTS bearers (i.e., between the mobile terminal and Core Network gateway) is 100 ms, while for Radio Access Bearers (i.e., between the mobile terminal and the Core Network edge node) it is 80 ms. ^[5]

After a short review of the mobile network channels for video telephony, in the next section we will describe the protocols and codecs standardized for circuit-switched and packed-switched video telephony.