Combined Dynamic Power- and QoS- Management in Embedded Video Surveillance Systems

Intelligent video surveillance (IVS) offers a large spectrum of different applications that have strict quality of service (QoS)-requirements. Beside high demands in computing performance, power efficiency is another major concern in these applications. For instance, recent IVS systems need to deliver compressed video data in high quality while using devices that are partly solar- or battery-powered. Since delivering high QoS-levels usually goes along with increased power consumption, it makes sense to implement combined QoS-management and power reduction methods. However, most IVS systems do have a lack of functionality in offering combined power- and QoS-management mechanisms. In this paper, we present PoQoS, a new generic concept for combined power- and QoS-management in distributed IVS systems. Its basic idea is to offer an extensible model that eases to implement coexisting dynamic power- and QoS-management mechanisms. Furthermore, our approach increases the systems reliability by including parameters of its actual energy state as well as information about fault recognition. Experimental results demonstrate the feasibility of our approach.