V.I. Parfenov1, Le Van Dong2

1,2 Voronezh State University (Voronezh, Russia)

1 Voronezh Institute of Russian Ministry of Internal Affairs (Voronezh, Russia)

Currently, as part of the development of radio engineering systems for various purposes, much attention has been paid to developing and applying innovative technologies for miniature computing and communication systems, including wireless sensor systems (WSNs). In particular, the simultaneous using several sensors in the WSN allows us to effectively solve the problems of detecting objects of interest or phenomena basis on their joint processing taking into account possible errors in such systems. Such errors are inevitable due to the presence of fading, environmental noise, etc., both at the level of local sensors and in the communication channel. In addition, they also depend on the rules functioning of both the sensors themselves and a fusion center (FC). When solving the problem of detection in the WSN, each sensor makes its local decisions about the presence or absence of an object, then transmits this information through a communication channel to a FC in which the received data’s joint processing is implemented and the final decision is based on the decision rule. Note that if in sensors decision scheme is used to make local decisions (Hard Decisions, HD), then the transmission such hard decisions needs less bandwidth and reduces the system’s communication costs. However, it loses a lot of sensing data information and restricts the improvement of the system’s detection performance. To improve the system’s detection performance in comparison to detection algorithms that use only hard local decisions, distributed detection algorithms based on using of the soft decision (SD) scheme in the local sensors are synthesized. As a result, the decision rule based on only soft decisions (LRT-SD) and the decision rule based on soft-hard decisions (LRT-SHD) are presented. The synthesized algorithm is based on the classical rule of detection, which is to compare the likelihood ratio with the detection threshold. The performance indicator is the total error probability. The detection threshold was determined numerically based on the criterion of an ideal observer. The obtained recurrent expressions for the probabilities of errors of the first and second kind allow us to accurately calculate the minimum value of the total error probability of the entire system. For the presented algorithms, the dependencies of the total error probability on the energy parameter taking into account the signal-to-noise ratio (SNR) at the level of local sensors and the channel’s signal-to-noise ratio, is given. Analysis of the calculated graphic dependencies allows us to make some conclusions and formulate recommendations for the construction of wireless sensor networks when their functioning. In particular, it is shown that under the considered conditions the LRT-SD and LRT-SHD detection algorithms have a certain gain in efficiency relative to the earlier known algorithm based on hard decisions (LRT-HD). Moreover, this gain is significant for sufficiently large values SNR in the system. In a non- ideal communication channel, the influence of interference significantly impairs the detection performance of the prestened algorithms. It is obvious that at a small channel SNR not exceeding 2, the total probability of entire system becomes unsatisfactory for small numbers of sensors (K=3,4). In addition, the system's detection performance also is improved by increasing the number of sensors. With restrictions bandwidth of the communication channel, it is advisable to use the LRT-SHD algorithm to solve the problem of object detection.

Parfenov V.I., Le Van Dong. Distributed detection algorithms based on using the soft decision scheme in local sensors. Radiotekhnika. 2022. V. 86. № 3. P. 14−26. DOI: https://doi.org/10.18127/j00338486-202203-02 (In Russian)

1. Sohraby K., Minoli D., Znati T. Wireless sensor networks: Technology, protocols, and applications. Hoboken, New Jersey: John Wiley & Sons, Inc. 2007. 236 p.
2. Dargie W., Poellabauer C. Fundamentals of wireless sensor networks: Theory and practice. John Wiley & Sons, Ltd. 2010. 330 p.
3. Parfenov V.I., Le Van Dong. Algoritmy kompleksirovanija informacii v besprovodnyh sensornyh setjah s uchetom verojatnosti vyhoda sensorov iz stroja. Radiotehnika. 2019. № 12(19). S. 53–59 (In Russian).
4. Parfenov V.I., Le Van Dong. Optimal'nyj algoritm kompleksirovanija informacii v besprovodnyh sensornyh setjah s uchetom vlijanija pomeh v kanale radiosvjazi. Telekommunikacii. 2020. № 2. S. 12–17 (In Russian).
5. Parfenov V.I., Le Van Dong. Analiz vlijanija okruzhajushhej sredy na jeffektivnost' algoritma obrabotki informacii v besprovodnyh sensornyh setjah. Fizika volnovyh processov i radiotehnicheskie sistemy. 2020. T. 23. № 2. S. 49–54 (In Russian).
6. Liang J., Hu Y., Liu H., Mao C. Fuzzy clustering in radar sensor networks for target detection. Ad Hoc Netw. 2017. V. 58. P. 150–159.
7. Junhai Luo, Xiaoting He. A Soft–Hard Combination Decision Fusion Scheme for a Clustered Distributed Detection System with Multiple Sensors. Sensors. 2018. V. 18. № 12: 4370. P. 1–18.
8. Thomopoulos S.C.A., Viswanathan R., Bougoulias D.C. Optimal decision fusion in multiple sensor systems. IEEE Transactions on Aerospace and Electronic Systems. 1987. V. 23. № 5. P. 644–653.
9. Lee C.C., Chao J.J. Optimum Local Decision Space Partitioning for Distributed Detection. IEEE Trans. Aerosp. Electron. Syst. 1989. V. 25. № 4. P. 536–544.
10. Skljar B. Cifrovaja svjaz': Teoreticheskie osnovy i prakticheskoe primenenie. M.: Vil'jams. 2003. 1104 s.  (In Russian).