Performance limits of multiaccess tree protocols with shared multi-user coding

Group random access methods usually involve apportionment of channel resources either in the time or the frequency domain, among a number of groups in a total population of users. Each group collectively accesses its apportioned resource according to some random access protocol. In the paper the author poses and answers the question whether a similar group-sharing can be accomplished in the code domain. He shows, using results in the theory of multi-user codes, that an M-user code can be effectively shared among M classes of users, rather than among M individual users. He analyzes the performance gains in the specific cases of Aloha and tree protocols for random multiple access.<
>