Multi-Objective and Financial Portfolio Optimization of p-Persistent Carrier Sense Multiple Access Protocols with Multi-Packet Reception
Ref: CISTER-TR-150103 Publication Date: 2015
Multi-Objective and Financial Portfolio Optimization of p-Persistent Carrier Sense Multiple Access Protocols with Multi-Packet ReceptionRef: CISTER-TR-150103 Publication Date: 2015
This paper revisits the study of wireless carrier-sense multiple access (CSMA) protocols enabled with multi-packet reception (MPR) capabilities using a new paradigm based on multi-objective and financial portfolio optimization. In this new paradigm, each packet transmission is regarded not only as a network resource, but also as a financial asset with different values of return and risk (or variance of the return). The objective of this network-financial optimization is to find the transmission probability of each terminal that simultaneously optimizes network metrics such as throughput and power consumption as well as economic metrics such as fairness, return and risk. Two models are considered for performance evaluation: a Bernoulli transmission model that facilitates analytic derivations, and a Markov model that considers the backlog states of the network and that allows for a dynamic stability analysis. This work is focused on the characterization of the boundary (envelope) or the Pareto optimal front curve (surface) of different types of trade-off performance region: the conventional throughput and stability regions, as well as new trade-off regions such as sum-throughput vs. fairness, sum-throughput vs. power consumption, and return vs. risk. Fairness is evaluated by means of the Gini-index, which is used in the field of economics to measure population income inequality. Transmit power is directly linked to the global transmission attempt rate. In scenarios with weak MPR capabilities, the system has problems in achieving simultaneously good values of fairness and high values of sum-throughput, mainly because of an underlying non-convex throughput region, which is typical of protocols dominated by unresolvable collisions. On the contrary, in scenarios with strong MPR capabilities, good fairness and high sum-throughput performances can be simultaneously achieved at the expense of power consumption. Carrier-sensing is shown to improve the convexity of the throughput region in scenarios with weak MPR, thereby achieving a better trade-off between metrics, including return and risk. However, the effects of carrier-sensing disappear in scenarios with strong MPR capabilities or with underlying convex throughput regions. The combination of MPR with carrier-sensing tools helps in reducing risk in the network and to fight issues of wireless random access such as the hidden/exposed terminal problems.
Chapter in book Communications in Computer and Information Science, Optimization in the Natural Sciences, Edited: Alexander Plakhov, Tatiana Tchemisova, Adelaide Freitas, Volume 499, pp 68-94.