Dror Baron - ALOHA with delay constraints Satellite application for ALOHA


My M.Sc. research considered ALOHA networks with a quality of service (QoS) constraint. A typical application might be a satellite hub that services ground-based stations. Because the delay in communication with the satellite is large, a QoS issue of practical interest is to minimize the probability of missing some delay constraint. In these systems, the round trip time is often much larger than the slot length, and delay constraints can be expressed in terms of round trips. This constraint changes the rules of the ALOHA game. Because traditional ALOHA retransmits once per round, and our constraints cap the number of possible retransmissions, the rates are severely degraded, especially when constraints are harsh. Another observation is that the probability of approaching the deadline is moderate, and so it makes sense to use lots of resources (retransmissions, lightly loaded ALOHA channels, etc.) as we approach the deadline. This leads to a judicious use of resources: few resources are used in initial rounds, while increasingly more resources are used as we approach the deadline.

The concept of using more resources near the deadline for delay-constrained ALOHA was introduced in a previous paper by Yitzhak (Tsahi) Birk (my M.Sc. advisor) and Yaron Keren. Their multi-copy approach used a monotone non-decreasing number of copies of the packet in each round up to the deadline. We considered different ways of allocating increasingly more resources as the deadline approaches.


Last updated December 2008