Smart Energy Management and Energy Distribution in Decentralized Self-Powered Sensor Networks using Artificial Intelligence Concepts

Zusammenfassung (Abstract) :

Sensorial materials equipped with embedded miniaturized smart sensors provide environmental information required for advanced machine and robotics applications. With increasing miniaturization and sensor-actuator density, decentralized self-supplied energy concepts and energy distribution architectures are preferred and required. Self-powerd sensor nodes collect energy from local sources, but can be supplied additionally by external energy sources. Nodes in a sensor network can use communication links to transfer energy, for example, optical links are capable of transferring energy using Laser- or LE diodes in conjunction with foto diodes on the destination side, with a data signal modulated on an energy supply signal. We propose and demonstrate a decentralized sensor network with nodes supplied by 1. energy collected from a local source, and 2. by energy collected from neighbour nodes using smart energy management. Nodes are arranged in a two-dimensional grid with connections to their four direct neighbours. Each node can store collected energy and distribute energy to neighbour nodes. Each autonomous node provides communication, data processing, and energy management. There is a focus on System-On-Chip design satisfying low-power and high-miniaturization requirements. Energy management is performed 1. for the control of local energy consumption, and 2. for collection and distribution of energy by using the data links to transfer energy. The loss of energy x (in the range between 0 and 1) occuring when â??energyâ? is routed along different nodes from a source to a destination node (assuming N intermediate nodes) reduces overall efficiency dramatically in the order of xN. By using electrical connections only neglible loss of energy can be expected in a distributed network, in contrast to optical and radio wave connections which have significant loss in the order of 10-30% per node. Additionally, in the latter case there is no physical interaction between a source and a sink requiring active management (routing). To overcome these limitations and to increase efficiency, smart energy management is performed by using concepts from artificial intelligence. Nodes of the sensor network can create groups (or one large group) and try to distribute and interchange energy based on local and global constraints and goals to be fullfilled. A mobile agent system is used to negotatiate energy demands and energy distribution and to implement group communication.