Efficiency or fairness: managing applications with different delay sensitivities in heterogeneous wireless networks

TL;DR

This paper proposes a pricing-based incentive scheme to improve the efficiency of user-network associations in heterogeneous wireless networks, accounting for different delay sensitivities, and demonstrates its effectiveness through analytical and simulation results.

Contribution

It introduces a simple, easily computable incentive mechanism tailored for heterogeneous users with varying delay sensitivities, enhancing network efficiency.

Findings

Properly designed incentives lead to efficient user-network associations.

The incentive scheme improves performance under dynamic demand conditions.

The scheme is simple to implement using readily available information.

Abstract

In the current intensively changing technological environment, wireless network operators try to manage the increase of global traffic, optimizing the use of the available resources. This involves associating each user to one of its reachable wireless networks; a decision that can be made on the user side, in which case inefficiencies stem from user selfishness. This paper aims at correcting that efficiency loss through the use of a one-dimensional incentive signal, interpreted as a price. While the so-called Pigovian taxes allow to deal with homogeneous users, we consider here two classes with different sensitivities to the Quality of Service, reflecting the dichotomy between delay-sensitive and delay-insensitive applications. We consider a geographic area covered by two wireless networks, among which users choose based on a trade-off between the quality of service and the price to pay. Using a non-atomic routing game model, we study analytically the case of constant demand levels. We show that when properly designed, the incentives elicit efficient user-network associations. Moreover, those optimal incentives can be simply computed by the wireless operator, using only some information that is easily available. Finally, the performance of the incentive scheme under dynamic demand (users opening and closing connections over time) are investigated through simulations, our incentive scheme also yielding significant improvements in that case.