I. Introduction
Wireless Sensor Networks (WSN) is the network comprised of a large number of nodes with sensing and routing capabilities [1]. The WSN have nodes with severe energy constraints, variable quality links, and low data-rate. Due to the availability of the low-cost cameras, microphones, and other sensors producing multimedia data such as images, audio, video have made it possible to gather information rich multimedia data from the physical world which will enhance the performance of target tracking, environmental monitoring, industrial process control and time critical applications, etc. in WSN and the resulting network can be termed as Wireless Multimedia Sensor Networks (WMSNs). The unique characteristics of multimedia data and design issues of WMSN are discussed in [2], [3]. The existing applications of WSN such as real time target tracking in battle environments, emergent event triggering in monitoring applications etc. are extended to video surveillance and notification, video and computer assistance in video-assisted living and healthcare to get improved performance. The stringent requirements of real-time multimedia applications include end-to-end delay; bandwidth and packet loss. Managing real-time data requires both energy efficiency and Quality of Service (QoS) assurance in order to ensure efficient usage of sensor resources and correctness of the collected information. Communication protocols for WMSN must therefore be specially designed to operate efficiently under these constraints. In order to satisfy the energy constraints and the QoS requirements for the WMSNs, clustering has been a common and active approach to organize sensor networks into clusters [4]. The range of the sensors' radio is in general quite short when compared to the network size of a typical WSN covering a large geographical area. Multi-hop routing is preferably used for the transport of the sensed data to sinks because most of the nodes would be far away from the sink and thus requires many hops to reach the sink [1] and as the path length becomes longer, the more energy is dissipated. As a result, response times become excessive and the lifetime of the WSN becomes very short. To save energy, a single-sink model is not scalable for large-scale WSNs. In fact, with increase in the number of sensor nodes, the information collected at the single sink some times might become excessive with respect to its communication capacity Recently, due to the scalability problems of single-sink network architectures made it to evolve towards scenarios with multiple sinks to achieve shorter paths and where nodes must form efficient data gathering trees and select the best sink to send these data [5], [6]. Multi-sink networks can remarkably reduce the mean distance between nodes and sink, basically resulting in energy saving and longer lifetime. The main focus in this paper is not only to develop energy efficient routing providing service differentiation for multimedia real-time data but to investigate use of multiple sinks in clustered WSN which improves the network lifetime. We compare the performance improvement using proposed protocol with multi-sink over the single sink model considering the impact on varying buffer size, minimum hop count, and minimum end to end delay.