Cross Layer Design and Optimization in Wireless Sensor Networks
MetadataVis full innførsel
This thesis targets some of the cross layer design issues presented in wireless sensor networks (WSNs). These issues will be analyzed and discussed from different perspectives in the following chapters. WSNs consist of many autonomous and self-organized sensor nodes, which operate in a predestined manner and are designed to achieve a certain application goal. WSNs are in many ways different from the common mobile networks. the most distinctive difference is that the sensor nodes are typically small and battery driven, which make them both energy and power limited; however, the nodes often are required to have a long operation time. Therefore, energy efficiency is crucial for WSNs. All these properties of WSN require large scale devoted research efforts. One of the most important features of WSNs is that multi-hop communications are often required to forward information from the source node to the sink node. As every node is energy constrained, then the optimal allocation of available resources along a multihop route, so that minimal energy is consumed and the data transmission quality is guaranteed, becomes a very crucial problem to consider. Several results and inspiring conclusions can be drawn through both theoretical analysis and computer simulations, which indicates that the energy can be reduced substantially compared to a non-optimized system. A few industrial standards targeting WSN applications have also been proposed, the most famous among which is perhaps the IEEE 802.15.4 standard. In this industrialized standard, cross-layer considerations have to certain extent been taken into account; however, the entire communication stacks are mainly based on the open system interconnection (OSI) structure which has clear separations between the medium access control (MAC) layer and the physical layer. One such example is the fact that carrier sense multiple access with collision avoidance (CSMA/CA) from IEEE 802.11 was selected as the standardized MAC protocol. This medium access protocol have been proven to work well in high traffic wireless local area networks (WLAN), but it does not take the physical channel into consideraition. Such negligence, particularly in the usually low-traffic WSNs, can result in very low energy efficiency. To overcome this drawback, we propose a new derivation of CSMA/CA, namely the so-called carrier sense multiple access with collision avoidance and channel adaptation (CSMA/CA2), as medium access method. The new scheme not only keeps the collision avoidance ability, which is an advantage of conventional CSMA/CA, but it also includes physical channel considerations. The simulation results show that the new protocol can substantially increase both the frame success rate and the energy efficiency. Lately, many researchers have also set their focus on possible potential WSN applications. One important direction is to use WSN for cognitive radio, and the main task for a WSN in cognitive radio applications is to continually sense the availability of the channel, and monitor the actions of the primary users so that an optimal policy can be formed to maximize the secondary users’ throughput. The latest research has shown that if the channel process can be described by a Markov process, the myopic policy turn out to be the optimal channel switching policy when the total number of channels is less than four. However, how the myopic policy behaves for more general statistical channel processes is still unknown. To deal with this, we discuss the performance of the myopic policy for a channel described by a general stationary, ergodic process. Simulations show that our analytical method provides an accurate tool for performance evaluations, and our proposed bounds on the achievable throughput can be shown to be very close to the exact performance.