Medium access control for multicast traffic in ad-hoc networks
23 September 2008
We study a variety of problems related to scheduling and medium access control for multicast traffic in wireless networks. Understanding this type of traffic is important because the broadcast nature of the wireless medium means that it is particularly suited to multicast applications. However, most work on contention-based and random-access scheduling focuses on the case of unicast traffic. In earlier work we looked at what types of multicast throughput are obtainable in networks that purely use random access. In this paper we examine the following issues.(i) Does it make sense to employ fanout-splitting when scheduling multicast traffic in a wireless network, i.e. does it makes sense to try and transmit to different receivers at different times? (ii) How much capacity is lost when we use random access as opposed to more coordinated schedules. (iii) Is it possible to do a Bianchi-style Markov chain analysis of multicast throughput in contention-based medium access control? What does this type of analysis tell us about the relative benefits of fanout-splitting vs non-fanout-splitting. Our results show that there are situations where fanout-splitting can increase throughput and that in networks of general topology the capacity loss due to random access scheduling can be unbounded. We also provide a Bianchi-style analysis for multicast traffic and examine the difference in throughput for the fanout-splitting case and the non-fanout-splitting case.