Buddies, not enemies: Fairness and performance in cellular offloading

Abstract

Recently, a number of offloading approaches have been proposed to reduce the burden on the cellular infrastructure, especially during peak hours. Ranging from pure data offloading concepts using local caches to fully-fledged services operating on ad hoc networks, these approaches are mostly tuned towards their performance. Consequently, cache hit ratios, achieved throughput, or end-to-end latencies, are prominent evaluation metrics. However, when utilizing users' resources to improve the performance of a shared service, the fairness with respect to the resources contributed by individual users should be considered as well. Resource consumption, e.g., the battery lifetime, can vary significantly between individual users, depending on their contributions. This effect can have a significant impact on user acceptance and, thus, the usability of the overall system. In this paper, we examine the trade-off between overall system performance and fairness w.r.t. resource utilization for a state-of-the-art monitoring service relying on offloading. To bridge the identified gap between performance and fairness, we propose a number of protocol adjustments to increase the system's fairness. Through an extensive simulation study we show that the proposed mechanisms lead to improvements in the overall achieved fairness of up to 60%, with below 4% degradation in service quality.