Abstract

Access to multimedia services is currently being considered by both wireless and wireline operators. Although satellites cannot compete directly with other alternatives, they can provide benefits in certain niche markets due to their wide coverage area and distance insensitivity. It is the goal of these satellite operators to simultaneously maximize channel utility (and revenue) and to meet the stringent quality of service requirements of the new multimedia services. To achieve both of these objectives, an efficient multiple access (MAC) protocol is required. Satellite MAC protocols have traditionally been based either on fixed assignment (for voice and video) and/or random assignment (for data). The unique feature of the newer multimedia services is that they generate a mix of these three traffic types, and moreover the traffic is much burstier than originally expected. As a result, the traditional protocols fail to provide the required performance. This work addressed this issue by proposing two new schemes--one based on combined free/demand assignment multiple access (CFDAMA) with a multi-frequency time-division multiple access (MF-TDMA) frame, and the other based on flow controlled random access with a multi-code time-division multiple access (MC-TDMA) frame. Performance results (for jitter-tolerant packet delay and real-time loss probability) show that both techniques perform well in a bent-pipe environment. For an on-board processing system, both protocols can be adapted to deal with possible congestion at the output of the on-board baseband switch. Performance of the modified CFDAMA scheme is superior to that of a rate-based technique proposed in the literature. For the proposed MC-TDMA system, soft blocking and congestion are both handled by controlling packet transmissions through a simple flow control parameter. Performance evaluation of satellite multimedia networks is also a very important topic. Event-driven simulations (with OPNET) prove invaluable to evaluate protocol performance, but these have limitations with regards to flexibility and accuracy (especially in evaluating real-time loss probability). As a result, certain key measures were obtained through analysis. In particular, the Matrix Geometric approach was applied to the CFDAMA scheme to determine real-time loss probability and jitter-tolerant packet delay.