Cooperative Caching in Wireless Multimedia Sensor Networks

Introduction

Wireless multimedia sensor network (WMSN) comprises of the large number of low cost scalar nodes and camera nodes that are deployed in hostile environment. WMSNs communicate over wireless medium and capture scalar as well as multimedia data. These nodes are battery driven devices (Akyldiz et.al. 2002), and it is difficult impossible to replace the batteries in hostile environment. WMSNs are resource constrained, so it is expected that these nodes perform their tasks for the maximum time period.

Maximum of WMSN applications are data centric, which transmit the sensed data to the sink node through single/multi hop transmission. The sink node is located close or inside/outside the sensing field. Sensor nodes sense the data from surrounding and transmit to the sink. Sink collects all data and performs some operation accordance with the application requirement and then transmits to the end user through the internet or some other wireless/wired media. Large numbers of nodes are deployed in sensing field. The large amount of multimedia data is transmitted and received by sink node that leads to higher energy consumption and short network lifetime of WMSN. To prolong the network lifetime, we have to perform less energy consumption tasks, i.e. less data transmission/reception operations are performed. This could be achieved if data communication to the sink is performed only as per the request of user/application via sink node. However, sink node can request data from any region at any time. Therefore, SN needs to perform continuous sensing to fulfill the objective and store these data items either in its own memory or by some other SNs memory in the network, so that data is provided to the sink node as and when the request comes. Since the storage capacity of SN is limited, it can store only the limited amount of data temporarily. However, cooperative caching is helpful for WMSN to maintain the large amount of data.

There are number of applications, such as capture an event, tracking the movement of an object, creating panoramic view of an area, etc. where sink node may require both recent as well as history data. There is another way to fullfill this requirement that sink node stores all history data, and all nodes have to transmit only the recent data to the sink. For this large amount of storage capacity is required at the sink, as the number of SNs continuously transmit sensed data to sink node. Although, many WSN protocols have been discussed in literature Manjeshwar et. al. 2002; Younis et. al. 2003; Poonia et. al. 2011; Yi et. al. 2007; Heinzelman et. al. 2002; Zhu et. al. 2009, it is assumed that sink node has ultimate power, capacity and storage capacity, whereas in other scenarios, it is assumed that sink node has limited resources as in simple SNs.

Figure 1 shows the query processing in wireless multimedia sensor network. An event has occurred, and it is sensed by number of SNs and they transmit data to the sink node. However, sink node requires both recent and history data, such as information regarding the event that took place at time t, where is the current time and is the history time interval for which data is required by the sink. Number of queries are generated by the sink node where query is generated at query is generated at and so on. Based on both recent and history data, sink node detects the occurrence of an event.