Stability and Delay of Network-Diversity Multiple Access with Backlog Retransmission Control
Ref: CISTER-TR-170402 Publication Date: 21 to 25, May, 2017
Stability and Delay of Network-Diversity Multiple Access with Backlog Retransmission ControlRef: CISTER-TR-170402 Publication Date: 21 to 25, May, 2017
This paper proposes two strategies for retransmission control of backlog traffic in the family of algorithms known as network diversity multiple access (NDMA). This type of algorithm can be relevant for future 5G systems, mainly because they provide (in ideal conditions) an almost collision-free performance for contention-based traffic, while achieving very low latency values with reduced feedback complexity. This matches the machine-type traffic, real-time, and dense object connectivity requirements in 5G. However, existing analysis generally ignores the backlog traffic generated by the imperfect detection conditions that arise in settings with finite SNR (signal-to-noise ratio). This paper aims to partially fill this gap, by providing analytic expressions for the performance of symmetrical training-based NDMA protocols with two different types of backlog traffic retransmission schemes. In the first strategy, all terminals involved in an unsuccessful resolution period retransmit immediately in the subsequent resolution periods or epoch slots. This procedure is repeated continuously (persistent retransmission) by inducing the same collision event under different channel outcomes until all the contending signals are correctly received. In the second retransmission strategy, the terminals in backlog state retransmit at a randomly selected time slot. In both strategies, expressions are here obtained on the maximum stable throughput and the delay experienced by any packet to be correctly received by the destination. This allows us to determine the capabilities of NDMA for achieving low-latency, reduced feedback complexity, as well as highly stable and real-time throughout performance.
Accepted in The Second International Conference on Advances in Signal, Image and Video Processing - from Sensing to Applications (SIGNAL 2017), 5G.