WIRELESS SENSOR NETWORK BASED BUILDING AUTOMATION SYSTEM

Smart grid adds intelligence and bidirectional communication capabilities to today’s power grid, enabling utilities to provide real-time-pricing (RTP) information to their customers via smart meters. This facilitates customer participation in demand response programs to reduce peak electricity demand. ZigBee, as a wireless technology characterized by low cost, power, data rate, and complexity, is ideal for smart grid applications. Unfortunately, almost all ZigBee channels overlap with Wireless Local Area Networks (WLAN) channels, resulting in severe performance degradation. In this dissertation, firstly, practical ZigBee deployment guidelines under WLAN interference are presented. “Safe Distance” and “Safe Offset Frequency” are identified by means of a comprehensive approach consisting of theoretical analysis, software simulation, and empirical measurement. In addition, a frequency-agility based interference avoidance algorithm is proposed and is implemented. The proposed algorithm detects interference and adaptively switches nodes to “safe” channels dynamically to avoid interference with low latency and energy overhead. Later we propose a real-time residential demand response scheme based on price information provided by ZigBee enabled smart meters. We model the price signal as a random process with known deterministic components, and formulate the problem of cost-minimization as a stochastic scheduling problem. A constraint on the total power consumption is considered to avoid peak shifting. An opportunistic scheduling algorithm consisting of two stages is proposed. Each user is tentatively scheduled based on the optimal stopping approach as if there is no power constraint at the first stage, while the power constraint is taken into account at the second stage. The proposed scheduling algorithm can be implemented in either a centralized or distributed fashion. Simulations validate proposed scheme can evidently reduce the cost of electricity users. At last, an optimal and automatic residential load scheduling scheme with distributed storage system based on our price prediction scheme is proposed. The price prediction scheme forecasts the future 24-hour prices according to day-ahead pricing and RTP history. Our designed scheme adopts integer programming computation to provide simple solution for customers to minimize the total cost including electricity expenditure and storage degradation while satisfying customer time requirement. The performance of proposed scheme is then evaluated through simulation.Ph.D. in Electrical Engineering, July 201