Autonomic control for wireless sensor network surveillance applications
Darminder Singh Ghataoura, John Mitchell, George Matich · 2011
Distributed unattended ground sensor (UGS) networks are deployed to support surveillance capabilities, within a security-sensitive region. Distributed operation can be enhanced through situational awareness, which is a three level integrated approach, involving perception, comprehension and projection, concerning the monitored threat. In this paper, we present a temporal projection process concerning the monitored threat, in order to allow efficient consumption of network resources through autonomic transmission control. Temporal projection can be formalized by utilizing a Partially Observable Markov Decision Process (POMDP). The POMDP can be further enhanced to achieve savings in UGS network resource consumption, by managing when decision epochs concerning transmission selection and priority scheduling should be taken. Rather than to take these decision epochs on every threat observation, we propose it is more efficient to adapt decision epoch frequencies according to monitored threat characteristics. In addition to the POMDP we also present strategies for managing decision epoch frequencies for autonomic transmission control decision making, according to the characteristics of the monitored threat. Simulations have been conducted to verify the network resource savings that can be achieved, while evaluating the surveillance utility provided, concerning the monitored threat, through using our proposed decision epoch interval strategies.