This paper proposes a scheme for updating the location of the sink node to balance the network topology when a wireless sensor network (WSN) is scaled up. We divide the proposed location update scheme into two steps, namely, searching the optimal location and designing the pathfinding algorithm. For the former, to find the optimal location of the sink node simply and efficiently, we only consider the information of the expanded longer paths and some key nodes instead of the global information of the entire network, which is easy to implement with a low-computational load. Then, considering the general unattended application scenario, we propose an improved reinforcement learning (RL) algorithm for the sink node to calculate a feasible efficient path, and then the sink node follows the path to reach the optimal location. Finally, through simulations, we demonstrate the optimal position of the sink node in expanded scenarios and successfully let the sink node learn the effective pathfinding method to reach the target position. A large number of simulation results verify the efficiency and effectiveness of our proposed scheme from the perspective of the efficiency of the pathfinding algorithm.
We propose a novel signal multiplexing and detection method for multiple-input multipleoutput (MIMO) communication systems, especially when the number of transmitting and receiving antennas is limited. Inspired by the idea of Compressive Sensing (CS) which can recover a given signal vector from a vector of measurements with less dimensions, our proposed CS-based multiplexing scheme can deliver a modulated data vector with length l via a MIMO system with fewer transmitting/receiving antennas than l, offering higher multiplexing gain. On the receiving side, our proposed detection scheme has two steps, which resort the BCS algorithm and a Deep-Learning algorithm to recover the original modulated data vector. Analytical and simulation results show that the proposed multiplexing and detection method can achieve larger multiplexing gain while reserving good bit error rate (BER), offering a novel research paradigm to improve the utility rate of multiple antennas.
Data aggregation is one of the most important functions provided by wireless sensor networks (WSNs). Among a variety of data aggregation schemes, the coding-based approaches (such as Compressive sensing (CS) and other similar programs) can significantly reduce traffic quantity by encoding the raw sensed data using weight vectors. The critical feature to design a coding-based data aggregation protocol is to construct a weight/measurement matrix for the application scenario. After that, the sink node assigns the column of the matrix, which is treated as the weight vector during the encoding process, to each sensor node respectively. However, for a dynamic scenario where the number of sensor nodes changes frequently, the existing approaches have to reconfigure the network by regenerating the measurement matrix and allocating the new weight vectors for all the existing nodes, which causes a considerable energy consumption and affects the regular monitoring tasks. To solve this problem, we propose a Vandermonde matrix-based scalable data aggregation protocol (VSDA), which preserves the advantages of coding-based schemes and addresses the issues mentioned above. In VSDA, as new nodes join into the scaled-up network, the original weight vectors owned by the original nodes do not need to regenerate the weight vectors entirely but add some new entries by itself at all. It outperforms the existing schemes by saving the energy in network scaling-up. Besides, we propose a concise hardware framework to quantify the data encoding process of VSDA, which provides a performance analysis process that is closer to practical application. The numeric tests validate the performance of VSDA compared with the existing schemes in several aspects, such as, the number of transmissions, energy consumption, and storage space showing the outperformance of VSDA scheme. INDEX TERMS Wireless sensor network, data aggregation, vandermonde matrix, measurement matrix.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.