Optimizing Sensor Identification in Long-delay Networks to Account for Maximum Frame Size and Variations in Propagation Speed

Abstract

Long-delay networks (LDNs) are networks in which the propagation wave speed is lower than that of radio waves, such as in underwater acoustic networks (UANs). The number of nodes is normally large in such a sensor network and the number may very due to different factors such as power failure or environmental disasters. An identification procedure is needed in this network to observe which nodes are currently operational and a large amount of time can be wasted in every probe of the procedure due to the long propagation delay. Optimizing the number of probes improves the identification time and power consumption by 75% and 60% respectively in both the slotted and un-slotted cases [1]. While optimizing the number of probes for the long delay, the frame size (the time within which the nodes send their packets) increases due to the lower offered load. In this work we show that even with the limitation of the maximum frame size our procedure works well. One of the limitations of LDNs is the variation of the propagating wave speed. We observe that if the standard deviation of the propagation speed is approximately less than 1/e of the packet size then the identification procedure for the slotted case is better than that for the un-slotted case. In order to alleviate the effect of variation in propagation speed we use a guard time in the slotted case.