Performance analysis and power control of a concatenated coded multi-media DS-CDMA system

Direct-Sequence Code-Division-Multiple-Access (DS-CDMA) is widely used in mobile communications for real-time voice transmission and can support naturally multi-rate traffic with different performance requirements by using different spreading factors and/or different transmit power levels. Packetized data can also be supported when Automatic-Repeat-reQuest (ARQ) scheme is used. However, in order to support multi-rate and multi-media traffic, many design issues need to be studied in details. We study a DS-CDMA system with concatenated Forward-Error-Control (FEC) code and perfect power control. Analytical formulae for the system throughput, packet drop rate and delay are obtained as a function of the code parameters for the inner and the outer codes. Using these analytical results, we study and optimize various system parameters for systems with homogeneous (same rate and type) as well as non-homogeneous users (different rates or types). For systems with homogeneous users, we consider both systems with only real-time users with no retransmission and systems with only non-real-time users with ARQ. For systems with real-time users, we study the effect of different codes and spreading factors on the system throughput under a fixed packet drop rate requirement. We also study the issues of call admission strategy and effects due to power control inaccuracy. For systems with non-real-time traffic, we study the effect of different codes, spreading factors, and retransmission probabilities on the system throughput under a fixed packet delay requirement. For systems with non-homogeneous users, we consider a system with only real-time users but different transmission rates and a system with both real-time and non-real- time users together. We study the power control strategy for these systems and find that the performance of DS-CDMA systems is very sensitive to the power control strategy. For the systems investigated, the Quality-of-Service (QoS) for real-time and non-real-time traffic are different. The supported traffic intensity is limited by the users with the higher QoS requirement, and reducing the transmit power of the users with the lower QoS requirement can raise the throughput of all users. Bit Singal-to-Noise-and- Interference-Ratio (SNIR) for every user is a parameter that can be tuned, but we found that proper tuning of another power control parameter, the maximum transmit power, can have a much more significant effect