This project examines wireless communication networks whose nodes have batteries that recharge by harvesting energy from the environment. Energy harvesting can yield a network that lasts as long as the network's hardware and intended purpose remain viable; this may be arbitrarily longer than the lifetime of any suitable single-charge battery. While average-power minimization is adequate to describe the lifetime of a device with a single-charge battery, a complete characterization of a network of rechargeable devices will depend on how the batteries are replenished. Given the broad variety of energy recharging systems, including solar cells, vibration absorption devices, wind and water mills, and thermoelectric generators, battery recharging is modeled as an environmental stochastic process. This project applies analytical models for battery recharging to evaluate fundamental multiple access, broadcast and relay network models composed of rechargeable nodes. The project objective is an enhanced understanding of the analytic fundamentals of rechargeable networks in order to contribute to the development and ultimate deployment of ecologically-friendly rechargeable networks.
This project examined networks whose nodes have batteries that recharge by harvesting energy from the environment. Energy harvesting can yield a network that lasts as long as the network's hardware and intended purpose remain viable; this may be arbitrarily longer than the lifetime of any suitable single-charge battery. Energy recharging devices may include solar cells, vibration absorption devices, wind and water mills, thermoelectric generators, and microbial fuelcells. Given this broad variety of existing devices and the high likelihood of the emergence of new devices, this three university collaborative project modeled energy recharging as an environmental random process and we analyzed and designed communication and network protocols for nodes subject to this random recharging. At Rutgers, the focus of this project has been the development and evaluation of models for energy harvesting receivers for coded packet communication. Models at two timescales have been developed. Both models share a modular partitioning of the receiver into (1) a sampler that represents the receiver front-end processing that transforms the received signal into symbol samples, and (2) a decoder that infers what codeword was transmitted from the sampled signal. While both functions are essential, the key idea in this partition is that sampling must occur in real-time while a packet is being transmitted. By contrast, once the signal samples have been collected and stored, decoding can occur at any time in the future. The intellectual merit of this project comes from the following project outcomes: A receiver energy model that separates real-time and offline processing. A coordinated variable-timing packet protoocol in which the transmitter sends a packet when the receiver has the energy resources needed to receive and decode a packet. For symbol-by symbol operation, the variable timing protocol is shown to achieve an energy conservation upper bound on the maximum reliable communication rate. A "deferred decoding" receiver protocol in which the receiver decodes packets using energy resources that would otherwise be discarded. For battery-constrained devices, this protocol is shown to be optimal for packet-by-packet operation. The broader impact of this work is that the improved understanding of the analytic fundamentals of rechargeable networks will contribute to the development and ultimate deployment of ecologically-friendly rechargeable networks.
