Mechanisms for energy efficient scheduling-based MAC for wireless sensor networks.

摘要

Energy efficient MAC design is a critical objective for wireless sensor networks which are usually highly energy constrained. In addition, the throughput and latency performance is also important for several sensor network applications. Several wireless sensor networks applications are characterized by the many-to-one communications; from sources to sink; instead of the one-to-one mode in ordinary ad-hoc networks. This type of communications might cause congestion at nodes close to the sink. Therefore, congestion avoidance should be used. Reducing data redundancy due to spatial and temporal correlation in sensing data is also an important goal. Further more, major energy waste factors in wireless communications, i.e., idle listening, overhearing, collisions and control overhead, influence the design of MAC protocols. Given these application and protocol design considerations, this dissertation proposes several mechanisms to optimize the energy consumption while maintaining high throughput and application specific latency.;OCSMACS's scheduling setup phase requires performing two main operations: neighbour discovery and topology collection. These operations are carried out by exchanging control messages using CSMA. For applications that require fast network deployment and/or in large or dense sensor networks, the use of CSMA results in contention, large number of collisions and subsequently more time is needed and more and more energy is consumed to collect all topology information. Therefore, we propose a time- and energy-efficient progressive topology construction protocol, called PROGRESSIVE, in which gradual topology information reaches the sink at the same time as TDMA slots are assigned to already discovered nodes. PROGRESSIVE controls the time during which CSMA is used, hence minimizes energy consumption. In addition, nodes can start data transmission as soon they are scheduled by the sink.;The third part of this dissertation is dedicated to exploiting the presence of correlation in the transmitted data to minimize energy consumption. The proposed spatial-temporal scheduling (TD and BU) assumes that each node that receives a data packet needs to forward it as it is. This means that during the data collection period the number of assigned transmit slots is at least equals to the number of receive slots. If data is highly correlated, then few slots are needed to transmit the aggregated data. However, TD and BU assign a considerably higher number of slots than needed. Forcing nodes to wake-up at specific time slots without transmitting or receiving data is a waste of energy and network resources. Therefore, we propose Correlation-Aware (CA) scheduling which allocates enough slots based on the level of data correlation. Two correlation models are considered: Global Correlation model and the proposed Sensing Range Correlation model. Simulation results show that OCSMACS together with the proposed correlation-aware scheduling improves energy efficiency and extends network lifetime beyond what is achievable using TD and BU scheduling.;To address the identified energy waste factors, we propose a new centralized multihop scheduling TDMA MAC protocol called On-Demand Convergecast Scheduling based MAC (OCSMACS). OCSMACS supports data collection (source-driven), event-driven and query driven WSN applications with the objective of energy efficiency and delay guarantee. In addition, it employes an adaptive on-demand slot assignment, compact and aggregate scheduling requests and integrated routing/MAC. These mechanisms allow nodes to sleep for long periods of time and wake-up at specific time slots to send and/or relay requests for slot assignment to the sink. Based on current topology information, the sink creates a multihop schedule that carries a new slot assignment for the requesting nodes as well as the relaying ones. OCSMACS relies on two proposed spatial-temporal multihop scheduling schemes: Top-Down (TD) and Bottom-Up (BU) scheduling. Both scheduling schemes facilitate the flow of data from nodes to the sink. In addition, they explicitly specify which slot is for send, receive or sleep. Simulations show that OCSMACS outperforms other well-known MAC protocols in terms of latency, throughput and energy efficiency. Furthermore, this performance comes at a scheduling cost that is diminished by the overall gain.