Skip to Main Content
Networks, including various types of Multi-Hop Wireless Networks (MHWN), such as Wireless Ad-Hoc Network (WAHN) and Wireless Mesh Networks (WMN) ubiquitously possess a user locality property, where adjacent users tend to exchange data with common locally popular destinations. Consequently, current congestion control mechanisms which are based on end-to-end signaling (e.g. TCP) do not make use of the aggregated congestion information and produce many redundant control signals. In this paper, we develop a new Collective Congestion Control (C3) paradigm for MHWNs. We first introduce a composition of the regional user nodes to an hierarchy of groups. Within each group we employ a new Collaborative Token Bucket (CTB) algorithm, which implements an effective distribution of the transmission rates and allowed data reception rates among users for communication with each common destination. Next, we demonstrate by simulations that our method yields a higher throughput and faster convergence to the desired performance level as well as better rate fairness compared to traditional end-to-end congestion control schemes. We show that the C3 implementation results with a much smaller amount of a control traffic which is critical in the MHWN resource limited environment. Finally, we analyze the CTB mechanism and demonstrate its advantages over multiple token buckets solutions.
Date of Conference: 3-5 Feb. 2010