Holistic Operations in Large-Scale Sensor Network Systems: A Probabilistic Peer-to-Peer Approach

Large-scale Sensor Networks Smart sensor nodes [14, 15] integrate multiple sensors (e.g., temperature, humidity, accelerometers), processing capability (microprocessor connected to the sensors using I2C technology), wireless communications, and a battery power source. Sensor nodes range around a few millimeters or smaller in size [11, 12]. Sensor nodes have low processing, memory and wireless communication capabilities. Processor and bus clock speeds of a few Mhzs, RAM sizes of a few KB, flash memory sizes of a few MB, and wireless bandwidths of a few hundred Kbps are typical [14, 15]. These ranges are often determined by application domain constraints such as cost, power consumption, and deployment space. Thus, they are likely to stay in spite of improvements in fabrication techniques. For example, retinal sensor sizes depend on ganglial nerve separation [11]. Space limits on the wireless antenna restricts maximum frequency and range of transmission. Availability of power is limited due to the difficulty or impossibility of recharging nodes in inhospitable terrains, e.g., military applications [12]. On the other hand, sensor nodes are becoming easier to produce in large numbers [14, 15]. An autonomous system consisting of a large number of sensor nodes deployed over an area and integrated to collaborate through a network (wireless or wired) network, would encourage several novel as well as existing applications. Examples include high fidelity image processing in retinal prosthesis chips [11], battlefield applications such as vehicle tracking, environmental observation and forecasting systems (EOFS) [13], etc.