Broadcast analysis in dense duty-cycle sensor networks

• Authors: S. K. Das;A. Di Saverio;G. Ghidini;A. Navarra;C. M. Pinotti
• Venue: Proceedings of the 6th International Conference on Ubiquitous Information Management and Communication
• Year: 2012

In this paper, the broadcast problem in dense Duty-Cycle Wireless Sensor Networks (DC-WSNs) is considered. Sensors are assumed to follow a duty-cycle schedule (d, L): they repeatedly alternate between an awake period of d time slots and an asleep period of L -- d time slots. A DC-WSN (n, r, δ), with n sensors uniformly and randomly deployed in a unit square area, transmission radius r for each sensor, and duty-cycle schedule ratio δ = d/L, is shown to be equivalent -- in terms of connectivity -- to a Random Geometric Graph RGG with the same number of sensors n and with a smaller sensor transmission radius r√δ. We consider a flooding algorithm that disseminates a message from a source to all the sensors in the network. Such an algorithm accomplishes the broadcast by means of a multi-hop communication schedule, without requiring a complete clock synchronization among sensors. By a large set of experiments, we verify that, if a DC-WSN(n, r, δ) is equivalent to a RGG graph connected, with high probability (w.h.p.), all the sensors receive the message when the flooding algorithm is performed. For evaluating the broadcast latency, for each sensor s, we count its distance from the source as the minimum number of times the message has be forwarded to reach s. We term the maximum among such distances as the network diameter. The network diameter depends on the maximum Euclidean distance in the deployment area (space latency) and on the inverse of the duty-cycle delay (time latency). The observed average diameter is minimum when 0.1 ≤ d ≤ 0.2. For such values of δ, it is proved analytically that, when d = 2, the space and the time latencies are balanced. Also the energy consumption and the broadcast latency are minimum when 0.1 ≤ δ ≤ 0.2 and d = 2.