Research On The Node Scheduling In Wireless Sensor Networks

Wireless sensor networks (WSNs) were identified by Business Week as one of the most important and impactive technologies for the 21st century. Wireless sensor networks are the product of the combination of computing, communications and sensor which are the three pillars of the information industry in the 21st century. The central premise of sensor networks is the distributed collection and digitization of data from a physical space, providing an interface between the physical and digital domains. Wireless sensor networks consist of a potentially large number of sensor modules that integrate memory, communication, processing, and sensing capabilities. The sensor modules form ad hoc networks in order to share the collected physical data and to provide this data to the network user or operator. Sensor networks have a wide range of applications, including medical, environmental, military, industrial, and commercial applications.Layered communication approaches typically separate communication tasks into several layers, with a clear definition of the functionality of each layer. In a layered communication stack, interaction among layers occurs through well-defined standard-ized interfaces that connect only the neighboring layers in the stack. In contrast, cross layer approaches attempt to exploit a richer interaction among communication layers to achieve performance gains. It led researchers to consider cross layer design for wireless sensor networks.With the development of wireless technologies, multifarious standards are cur-rently emerging. For example, the unlicensed 2.45 GHz ISM band can host various net-works with different standards, such as IEEE 802.11g WLANs, IEEE 802.16e WMANs and IEEE 802.15.4 WPANs. Thus, the coexistence issue of such networks challenges the reasonable and efficient use of the scarce spectrum. Fortunately, cognitive radios have been proposed as a technology to implement opportunistic sharing. They are able to sense the spectral environment over a wide frequency band and use this information to opportunistically provide wireless links that can satisfy the user communications requirements optimally.As a special kind of Ad-hoc networks, wireless sensor networks have some different characteristics such as:Sensor nodes are densely deployed. The data collected by the sensor nodes are highly correlated and redundant.The lifetime of the wireless sensor networks is highly dependent on the energy efficiency and fairness among sensor nodes.Considering the above characteristics of wireless sensor networks, we introduce cross layer design and cognitive radio technique, focus on the node and resource scheduling, to obtain better energy efficiency and fairness.The main work and the innovations are as the following:1. Considering the characteristics of wireless sensor networks and their applying sce-narios, we introduce the design challenges and guidelines for wireless sensor network cross layer design. We provide an overview of the features of existing cross layer approaches for wireless sensor networks that rely on information sharing and design coupling. The classification is by the features such as input aspect, configuration optimizations and performance goals.2. Noting that the channel state information (CSI) between the cluster head and the sensor nodes will affect the received bit energy noise ratio of the sensor nodes, we propose an optimal data fusion algorithm based on CSI for a one-hop clustered wireless sensor network. The cluster head receives the information bits from the sensor nodes for the final decision-making. The ones with good channel state should play more important role in the final decision-making. On the other hand, the ones, which have higher error probability, may have opposite effect on the fusion. So the fusion could not rely on them very much or even ignore them. On the basis of the fusion algorithm, we consider the redundancy of the sensor deployment and propose a cross layer transmission scheduling scheme. By selecting proper set of sensor nodes to transmit their local decision back in turn, the scheme can prolong the lifetime of the sensor network. The numerical and simulation results show that it can get a good tradeoff between the energy efficiency and the performance. 3. Considering the shortcomings of centralized control and no energy consumption consideration for antenna selection in the existing clustered virtual MIMO trans-mission mechanism, we propose a new virtual MIMO transmission scheme based on the directed diffusion routing protocol. The operations of the proposed scheme are broken into rounds. In each round, firstly the sink node disseminates the inter-est packets throughout the sensor network. These interest packets dissemination are used for setting up gradients within the network. Then the sensor nodes se-lect appropriate nodes to take part in the transmission or reception according to the gradients, finally complete the process of data sending back. This distributed scheme combines the MIMO technique on the physical layer and the directed dif-fusion protocol on the routing layer together. It can improve the energy efficiency, reliability and scalability of the networks.4. We consider the coexistence of wireless sensor networks with other wireless net-works using cognitive radio technique. Multiple sensor nodes are involved in the spectrum sensing to avoid the interferences from other wireless users. The more sensor nodes cooperate in the sensing, the better detection performance can be obtained, however, more energy is consumed. How to get the tradeoff between the energy efficiency and the detection performance is a key problem. We first obtain the least required detection time of a single sensor node when given the requirements on detection. Then, the voting fusion rule is adopted for the final decision making, the relationship between the final detection performance and the energy consumption is analyzed. Based on the considerations above, a detection scheduling matrix is presented in order to make the cooperative sensing more fairly. The cooperative sensing scheduled by the matrix can achieve a balance of energy consumption among the cooperative sensor nodes.5. We study the problem of optimally placing sensing times over a time window so as to get the best estimate on the parameters of an on-off renewal channel for the wireless cognitive sensor networks. We demonstrate that when sampling is done sparsely, random sensing significantly outperforms uniform sensing. In the special case of exponentially distributed ON/OFF durations, we derive tight lower and upper bounds on the Fisher information under a sparsity condition, while obtaining the best and worst possible sampling schemes measured by the Fisher information. We show that uniform sensing is the worst one can do; any deviation from it improves the estimation accuracy. We present a dynamic programming approach to obtain the best and worst sampling sequences in the more general case without the sparsity condition. We show that under the same channel statistics and the same average sampling interval (or frequency), a random sensing scheme affects the estimation accuracy through the higher-order central moments of the sampling intervals, and use the circularβensemble to study a family of distributions. We present an adaptive random sensing scheme that can very effectively track time-varying channel parameters, and is shown to outperform its counterpart using uniform sensing.6. The basic but the most important theory guidance for wireless sensor networks is the theory of information acquisition. However, the theory of information ac-quisition, storage, integration, processing and transmission under the guidance of Nyquist theory become the main bottlenecks for further development. So a new theory guidance, that is compressive sensing or compressive sampling is proposed. The theory demonstrates that:the signal can be compressed far below the Nyquist sampling rate in a standard way, still be able to accurately recovered. It is tempting to examine whether this technique brings any advantage for our channel estima-tion problem. The idea is to randomly sample the channel state, use compressive sensing techniques to reconstruct the entire sequence of channel state evolution, and then use the ML estimator to determine the channel parameter. While the compressive sensing technique can result in a whole sequence, its estimation per-formance is ultimately fairly poor due to the unsatisfactory reconstruction process. This points to an interesting direction of future study, which is to find a better basis matrix that can both sparsify signal and at the same time is sufficiently incoherent with the measurement matrix.