Overview of Voice Capacity

It is interesting to note that it is possible to run a Voice over IP service over the interactive bearer available in all existing deployments of UMTS. As the link level retransmissions are performed from the RNC with the RLC protocol (see Section 5.6) it is not possible to maintain the same delay (or quality) as for CS voice service, and the lack of header compression (and other system optimizations) makes the service much less efficient than the CS voice. For those reasons we will not study in detail the performance of the VoIP service over interactive bearers, but will rather focus on what is technically possible with the conversational PS bearer described in Section 5.6.

In Figure 7.6 we show an overview of the voice capacity potential for current and evolved UMTS access over both unoptimized and optimized conversational PS bearers.

Starting with original dedicated channels without header compression or any voice specific system enhancement it is possible to implement a VoIP service with quality equal to or better than the CS voice. However, service realization will consume substantially more resources than providing voice service over the CS bearers. This is shown as the leftmost bar in Figure 7.6.

It is possible to increase the capacity simply by using the HSPA, which will increase the capacity for all services. However, the resulting system capacity is still far below the CS voice capacity. The main reason for this is the header overhead introduced by the RTP, UDP and IP headers, and implementing header compression will increase the capacity close to the CS capacity.

Both HSPA and header compression are already standardized in 3GPP, and will also benefit services other than VoIP. Therefore they can be implemented in a system without a specific VoIP optimization. Together they will bring the VoIP capacity closer to the CS capacity, but will not match the CS capacity.

Other enhancements allow VoIP capacity to exceed the CS capacity, but are much more closely tied to the properties of the voice service. For example, it is possible to boost the capacity significantly by using a scheduler optimized for voice traffic ('delay scheduler') or by optimizing the radio access bearer for the used voice service and codec as described in Section 5.6. These two improvements allow VoIP over HSPA to exceed CS capacity,

DCH 64, no ROHC

HSPA, no ROHC

HSPA, with ROHC

+ improved scheduler

+ efficient signaling & protocols

HSPA Evolution LTE 5 MHz 5 MHz

Figure 7.6: The technology potential for IMS voice capacity. The five leftmost values are obtained with detailed system simulations. The sixth is a possible target for the HSPA Evolution, while the rightmost values are 3GPP requirements for Long Term Evolution.

DCH 64, no ROHC

HSPA, no ROHC

HSPA, with ROHC

+ improved scheduler

+ efficient signaling & protocols

HSPA Evolution LTE 5 MHz 5 MHz

Figure 7.6: The technology potential for IMS voice capacity. The five leftmost values are obtained with detailed system simulations. The sixth is a possible target for the HSPA Evolution, while the rightmost values are 3GPP requirements for Long Term Evolution.

but require tight coupling of the service and the radio access. We will study the impact of the scheduler in more detail.

Also shown in Figure 7.6 are the targets for the evolution of the radio access (see Section 3.3.2 for a description of different evolution paths). For Long Term Evolution, the agreed value is to support 200 voice calls in 5 MHz bandwidth, which corresponds to more than double the current CS capacity. There is no agreed target or goal value for the HSPA Evolution, but in general the performance should be similar as for LTE.

For HSPA evolution, the main improvements in the VoIP capacity is expected to come from the reduced overhead from control channels. For example, for HSDPA, it might be possible to reduce or eliminate the need to transmit the scheduling information on the HS-SCCH. This would not only save the power needed for HS-SCCH, but also allow easy scheduling of more than four users per TTI. For E-DCH, the current system needs to send power control feedback to control the downlink power. However, with HSDPA only downlink, the need for this power control is greatly reduced, and it might be possible to significantly reduce the amount of energy spent on the uplink control channels.

0 0

Post a comment