Design of intelligent surveillance system based on wireless ad-hoc network under special conditions

Abstract

For oil fields, shooting target range, military forbidden zones and other special complex scenarios, it is not only difficult to deploy communication infrastructure because of its remote location and difficult environment, but also difficult to transmit surveillance video images to command stations in real time through centralized communication networks. Furthermore, there is a waste of bandwidth resources and the use of wireless ad hoc networks to transmit surveillance video. Therefore, we aim at designing and implementing a real-time intelligent surveillance system based on wireless ad-hoc networks in special scenes in this project. The system is supposed to capture the video frames of the monitored area and perform automatic intrusion detection. Only when an intrusion is detected, the video frame is encoded and transmitted together with the alarm message to the command center. An improved Yolov2 algorithm to achieve real time object detection on a computationally limited platform was proposed first. A wireless ad-hoc communication system was designed and implemented on the basic of the Tactical Targeting Network Technology (TTNT) Data Link to transmit surveillance video stream, control signaling and alarm message. Then the extensive experiments were conducted to evaluate the system in terms of detection accuracy and communication performance. The experiments results demonstrate the effectiveness and accuracy of the proposed system, the size of the model is only 7.6% of Yolov2 and the detection speed is increased 4 times. This design transforms long-time data transmission into burst data transmission, thus saving bandwidth resources and improving bit error rate of 96.9%. It is of great significance for the future application of artificial intelligence and machine learning in the field of communication