Fairness in fairness: an interpretation and application in wireless ad-hoc networks

Fairness among all users is an important issue when accessing any shared-medium channel. For example, in a multi-hop wireless ad-hoc network, inherent conflict exists between achieving fairness and maximizing channel resource utilization. This conflict, coupled with the location dependent contention and the distributed nature of fair packet scheduling, makes it a very difficult task to even arrive at a clear definition of fairness in such networks. In this thesis, we explore the conflict in an attempt to define and achieve fairness using wireless ad-hoc network as an example. We propose several novel fair allocation algorithms that achieve fairness while exploring channel spatial reuse. We further devise corresponding localized version of the proposed algorithms. These allocation algorithms reside on top of distributed packet scheduling scheme to provide different levels of fairness within the IEEE 802.11 framework. Simulation of the distributed basic fair allocation algorithm is performed to demonstrate that it achieves the theoretical fairness objective with a higher throughput than IEEE 802.11 MAC. This is not only because of the lack of fairness consideration in IEEE 802.11 MAC, but also due to the fact that our packet scheduling scheme adjusts contention window in a more efficient manner. We also lay down the theoretical framework for investigating "fairness in fairness", which focuses on fair allocation of the bandwidth gain due to channel spatial reuse.

在接入任何共享介质信道时，所有用户之间的公平性是一个重要问题。例如，在多跳无线自组织网络中，实现公平性和最大化信道资源利用率之间存在内在冲突。这种冲突，加上与位置相关的竞争以及公平分组调度的分布式特性，使得在这类网络中甚至很难对公平性给出一个清晰的定义。在本论文中，我们以无线自组织网络为例，探讨这种冲突，试图定义并实现公平性。我们提出了几种新颖的公平分配算法，这些算法在探索信道空间复用的同时实现了公平性。我们还进一步设计了所提算法的相应本地化版本。这些分配算法位于分布式分组调度方案之上，以便在IEEE 802.11框架内提供不同程度的公平性。对分布式基本公平分配算法进行了仿真，结果表明它实现了理论上的公平目标，且吞吐量高于IEEE 802.11 MAC。这不仅是因为IEEE 802.11 MAC没有考虑公平性，还因为我们的分组调度方案以更高效的方式调整竞争窗口。我们还建立了研究“公平中的公平性”的理论框架，该框架侧重于对因信道空间复用而获得的带宽增益进行公平分配。