A Modular Energy Model for the Low Power Radio Domain in Linear-Topology Hybrid Technology Networking

The North American Freight Railroad industry has been exploring ways in which on-board real time or near real-time monitoring of important railcar components and cargo can be accomplished. This approach alleviates the danger from fast occurring catastrophic events like bearing failure, which is not always possible using the traditional wayside monitoring techniques. The use of Wireless Sensor Networks is a viable candidate technology that is being explored for this application. However, popular communication protocols based on IEEE 802.15.4 have been evaluated by the railroad industry and our lab, and were found to perform unacceptably for this application domain, among other reasons, as a consequence of the long linear chain-like network topology of a sensor network deployment on a train. Hybrid Technology Networking (HTN) protocol has been designed to address these issues. HTN structures the network as a communication hierarchy using multiple different network technologies. It allows small clusters to communicate internally using IEEE 802.15.4 and utilizes IEEE 802.11 as the inter-cluster transport method for data delivery over multiple hops to the locomotive. It aims to maximize the benefits afforded by each technology. Energy is a scarce resource in such networks and hence modeling it for energy analysis and optimization is vital. A model that accurately predicts the energy consumption of a particular network deployment is therefore of utmost necessity. However, most modelling efforts concentrate on network deployments utilizing only a single type of communication protocol and the structure of such deployments are often mesh-like. Also the existing modelling approaches tend to model the entire network as a single phenomenon, which is often not the case in network deployments such as those in the freight railroad scenario. It is also expensive to commission large network deployments to evaluate energy consumption profiles. The problem is compounded when this process has to be repeated for several different communication protocols and channel conditions. The task will be made economically viable and massively scalable with the use of a modular energy model. The philosophy behind our approach is to model the important and contributing constituents of the protocol within each node and also external to the node and then utilize inter-dependencies to connect the individual models. This work is the first step towards a modular energy model for the Hybrid Technology Network, and is also applicable to many other networking approaches. In this work we propose a model design that is capable of predicting the network behavior of nodes in a linear chain-like topology utilizing the ContikiMAC duty cycling protocol for multi-hop communication with the sink node. We have used channel emulation to test hardware nodes in a chain-like topology to validate the model predictions, and present our findings in this paper.