An open issue still to be addressed in low-power lossy networks (LLNs) is how the application requirements, the available transport services, the network layer routes, and the data link-layer resources are mapped efficiently. This can be explained by the fact that, in most LLNs, link-layer resources cannot be easily managed; this results in a best effort IP layer, and traffic engineering performed solely through flow control at the transport layer. The new IEEE802.15.4e standard defines a link-layer mechanism by which motes in the network synchronise and communicate by following a schedule. Each slot in that schedule can be seen as an atomic link-layer resource, which can be allocated to any arbitrary link in the network. The schedule can be built to match the bandwidth, latency and power requirements of each mote in the network. Managing that schedule is performed centrally in IEEE802.15.4e networks today. This paper explores a solution to achieve the same goal in a distributed manner. Specifically, we argue that this problem is very similar to traffic engineering on today's Internet.We show how multiprotocol label switching can be mapped to LLNs to manage the network's schedule. By using the completely fair distributed scheduler, we show by simulation how this novel link-layer resource allocation scheme yields a proper distribution of end-to-end delays among the motes and an average throughput that achieves the 70% of the maximum possible throughput in the worst conditions tested. Copyright © 2013 John Wiley & Sons, Ltd.
|Journal||European Transactions on Telecommunications|
|Publication status||Published - 1 Jan 2013|