An Information Value Driven Architecture for Urban Video Surveillance in Data and Attention Bandwidth Constrained Environments
Mahesh Saptharishi, Douglas Marman · 2009
Urban surveillance networks are characterized by network (data) and attention (security professionals monitoring copious sensors) bandwidth constraints. Both of these challenges exist because scarce network and attention resources are allocated under the assumption that all data are equally important. The proposed intelligent camera and video recorder (iCVR) architecture addresses the two bandwidth challenges directly by (a) enabling value assessment at the sensor level and (b) endowing the sensor with sufficient memory to store data for an extended time period. The latter is significant because the sensor, based on information value, can choose how long data need to be retained rather than just choosing to discard untransmitted data based on an arbitrary value threshold. We present the iCVRpsilas architecture and the impact of the value assessment and storage framework in the context of a wide-area urban surveillance application.