Secure distributed demand projection in micro-grids

Even more than in smart grid environments, the effective and reliable operation of micro-grids with limited ability to compensate for fluctuating demand by averaging over a large number of loads and relying on statistical models, requires micro-grids to rely on demand management. We argue that this can and should be augmented by explicit demand projection and to integrate the demand profile for loads into the balancing and demand management mechanism. For micro-grids in disadvantaged environments, this cannot rely on the integrity of loads or centralised building energy managers and smart meters; instead, demand projection and communication must rely on distributed, opportunistically employed devices with limited trustworthiness. This work employs a distributed algorithm based on approximate consensus problem over non-complete graphs using random geometric graphs, and describe an algorithm that enables a group of devices to agree on a way to use power that are approximately equal to each other in the presence of non-cooperating devices. In this paper we therefore describe a hierarchical model for power networks combined with hybrid communication networks for micro-grids together with algorithms allowing the timely communication of aggregate demand in the presence of non-cooperating entities and energy theft; for this our main contribution is to develop an efficient approximate consensus protocol for sparse, random geometric graphs.