Maximizing Growth Codes Utility in Large-Scale Wireless Sensor Networks

The goal of Growth Codes proposed by Karma et.al. is to increase the "persistence" of sensed data, so as to promise that data is more likely to reach a data sink. In many "zero-configuration" sensor networks, where the network topology would change very rapidly, Growth Codes are especially useful. However, the design of Growth Codes is based on two assumptions: (1) each sensor node contains only one single-snapshot of the monitored environment, and each packet contains only one sensed symbol; (2) all codewords have the same probability to be received by the sink. Obviously, these two assumptions do not hold in many practical scenarios of large-scale sensor networks, thus the performance of Growth Codes would be sub-optimal. In this paper, we generalize the scenarios to include multi-snapshot and less random encounters. By associating the decimal degree with the codewords, and by using priority broadcast to exchange codewords, we aim to achieve a better performance of Growth Codes over a wider range of sensor networks applications. The proposed approaches are described in detail by means of both analysis and simulations.