Efficient implementation of a dominance protocol for wireless medium access
Ricardo Gomes · Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT) · 2008
Embedded computing systems went through extraordinary evolutions during the past two decades, representing nowadays one of the most promising technologies for improving a wide range of application areas such as energy/resource management, safety, health or entertainment. New sensors and actuators are leading to an unprecedented level of interaction between computing systems and their surrounding physical environment. These embedded computers tend to be networked, often wirelessly, and they are becoming denser, of larger scale and more pervasively deployed. Since the wireless channel is a "natural resource" which must be shared between this large number of embedded computers, the medium access control (MAC) protocol significantly influences the performance of the entire system. In particular, satisfying real-time requirements — something that is needed for a computer to tightly interact with its physical environment — plays an important role. One solution was recently proposed by Pereira, Andersson and Tovar. It was a prioritized and collision-free MAC protocol belonging to a family of protocols called dominance/binary countdown protocols. This solution was implemented in commercial-off-the-shelf (COTS) wireless sensor networks (WSN) platforms and the implementation was demonstrated to be working. Unfortunately, those platforms had (for the MAC protocol) unfavourable characteristics which lead to limited efficiency and excessive overhead of the MAC protocol. This work presents a new hardware platform, in the form of a network adapter for common WSN platforms, that allows an efficient implementation of dominance protocols for wireless medium access, allowing the medium access to be performed in less than 5 ms for 216 priority levels, which represents an overhead reduction of more than ten times as compared to the protocol implementation in COTS WSN platforms. Additionally, the overall energy consumption was reduced by approximately 45 % when compared to the theoretical best-case performance of the protocol implementation in COTS WSN platforms. iv This work also allowed, for the first time ever, an aggregate computation scheme for WSN to work exploiting the new efficient implementation of a binary/dominance countdown protocol.