Modeling and evaluation of throughput, stability and coverage of RP-CDMA in wireless networks.

摘要

We revisit Random Packet Code Division Multiple Access, or RP-CDMA. Initially, RP-CDMA was investigated under the assumption that header transmissions are subject to Classical Aloha, where any overlap leads to the loss of all packets. Furthermore, the effects of multiuser interference were set aside and it was assumed, that the data frame could always be recovered perfectly as long as the header survives. No specific method for data detection was introduced or discussed. Also, unrealistically long data portions in the range of hundreds of thousands of bytes for the data frame have been assumed.; In this dissertation, we revise earlier results by adopting a more realistic Spread Aloha model for header transmission. There, header detection is subjected to chip-level collisions as well as multiuser interference. In addition, we discuss and evaluate the performance of various multiuser receivers for data recovery---namely the, matched filter, the decorrelation receiver, the minimum mean square error (MMSE) filter, a successive canceller as well as partitioned spreading demodulation (PS-CDMA). For the latter, we confirm and expand the understanding of its performance through Variance Evolution analysis. Moreover, we introduce two performance measures for multiple access systems based on the multipacket reception matrix E, termed Λ and zeta We show that A presents an upper limit on the vector of arrival rates that allow for stable network operation for a finite number of users. Hence, together with zeta the summation of the elements in Λ, the latter measure allows for a more useful classification of multiaccess systems than system throughput alone. We proceed to determine the system performance limits of RP-CDMA in base station centric networks under the assumption that all packets are received at equal as well as unequal power levels for the receiver methodologies mentioned above. To model data sizes, we resort to Internet2 traffic with upper packet sizes of 1500 bytes. We show that reliable and efficient network operation is feasible even with randomly chosen spreading sequences by deriving an analytical expression for the variance in per node capacity. Also, we motivate the idea that due to its unique properties, RP-CDMA can be used in Ad-hoc networks and present and discuss conclusions for such systems.