Model-driven performance engineering for wireless sensor networks with feature modeling and event calculus

Abstract

This paper proposes and evaluates a model-driven performance engineering framework for wireless sensor networks (WSNs). The proposed framework, called Moppet, is designed for application developers to rapidly implement WSN applications and estimate their performance. It leverages the notion of feature modeling so that it allows developers to graphically and intuitively specify features (e.g., functionalities and configuration policies) in their applications. It also validates a set of constraints among features and generates application code. Moppet also uses event calculus in order to estimate a WSN application's performance without generating its code nor running it on simulators and real networks. Currently, it can estimate power consumption and lifetime of each sensor node. Experimental results show that, in a small-scale WSN of 16 iMote nodes, Moppet's average performance estimation error is 8%. In a large-scale simulated WSN of 400 nodes, its average estimation error is 2%. Moppet scales well to the network size with respect to estimation accuracy. Moppet generates lightweight nesC code that can be deployed with TinyOS on resource-limited nodes. The current experimental results show that Moppet is well-applicable to implement biologically-inspired routing protocols such as pheromone-based gradient routing protocols and estimate their performance.