| Because of limited bandwidth availability and their typical dense, multi-hop deployment, wireless sensor networks have a fundamental need for efficient transport layer rate control. State of the art transport layer rate control protocols in wireless sensor networks are primarily heuristics that rely on Additive Increase Multiplicative Decrease (AIMD) mechanisms. This heuristic-based design of state of the art rate control protocols raises two key issues: first, due to the AIMD based mechanism the protocols suffer from long convergence times and large end-to-end packet delays; second, existing protocols are monolithic in design, either focusing purely on congestion control functionality without regard for rate utility optimization, or trying to optimize for a specific rate utility function. We improve upon the state of the art by proposing two rate control protocols that address the above issues. To address the issue of long convergence times and large end-to-end packet delays, we propose the Wireless Rate Control Protocol (WRCP). To address the issue of monolithic protocol design, we propose the Backpressure-based Rate Control Protocol (BRCP).;WRCP, to our knowledge, is the first explicit and precise rate control protocol for wireless sensor networks. WRCP has been designed using a novel interference model, called the receiver capacity model. The model helps determine the exact available capacity at each receiver in the network. WRCP uses the available capacity information presented by the model, and distributes this capacity amongst contending flows in a neighborhood in order to achieve a lexicographic max-min fair rate allocation. The use of explicit capacity information allows WRCP to exhibit fast convergence time. The explicit capacity information also allows WRCP to operate within the capacity region, resulting in small end-to-end delays.;BRCP is a, novel, flexible rate control protocol that has the ability to optimize for any concave rate-utility function. The design of BRCP is achieved by applying Lyapunov drift based stochastic optimization techniques to a Carrier Sense Medium Access (CSMA) based MAC. The ability of BRCP to make rate control decisions purely on local queue information makes it extremely useful in a wireless sensor network, since it requires minimal control information exchange. |
