PERFORMANCE COMPARISON OF QOS METRICS FOR A DISTRIBUTED PRICING SCHEME

De-centralized nature of nodes, in ad-hoc networks, results in the users adapting their operations independently. Such operations are mostly biased upon the figures and data available for the parameters which are imperative for superior performance or, in other words, improved Quality of Performance (QoS) of the nodes. In centrally controlled networks following cooperative game theory principles, collective operations are performed by the nodes for better QoS of the network. Although nodes in decentralized networks undertake individual operations, the final outcome of the whole network and thus the performance of the nodes in the network are influenced by the operations of other nodes. Hence, a distributed resource allocation approach in such a scenario can be modeled as a non-cooperative game. Asynchronous Distributed Pricing (ADP) is one such virtual pricing algorithm in which a user’s payoff is determined by the difference between how much a given performance metric is valued and how much is paid for it. User service demands and priorities are modeled using numerically emulated QoS metrics termed as utility functions. The network objective is to maximize the sum of all users’ payoff. However, for convergence of the sum of all users’ payoff to a global maximum, the determination of the QoS metric’s utility function with sufficient concavity is essential. Although supermodularity conditions have been previously defined and determined to obtain suitable utility functions, we have numerically and analytically illustrated that the convergence performance characteristics fluctuates with the choice of QoS metrics in the algorithm for similar utility functions as well. We have assessed the optimality of utility functions under Signal-toInterference-plus-Noise ratio and Signal-to-Interference ratio based calculations. This paper also explores into the difference in performance characteristics obtained by the addition of a significant value noise variance in the ADP algorithm