Our goal in this paper is to simulate the behavior of multiprocessor system on chip. We used an open virtual platform -OVPSim made by Imperas Company, which offers the possibility of programming and running application on the platforms architectures. With this platform we simulated both hardware architectures and running software applications. We used two types of processors -ARM7 IP core and MIPS32 IP core, shared memory, local memory and BUS for interconnections and simulated three systems on chip models and for each architecture we simulated the running of the same applications.I.

No abstract

Protected wetlands such as deltas, lakes or rivers provide a sanctuary for many endangered species. In order to protect these areas from illegal human interventions, it is necessary to monitor the unauthorized entrance of motor boats. In order to mitigate such an impact, we have developed a network of floating beacons for underwater acoustic monitoring, using LoRa communication modules operating at 433 MHz. Such beacons should be equipped with compact antennas. In this paper, we use a genetic algorithm approach to design the compact, monopole antennas required for the beacons; size constraints would apply not only to the radiating element but also to the ground plane. Although the antenna input is unbalanced, such a small ground plane may yield common mode currents on the antenna feeder, which distort the radiation pattern of the antenna. In order to investigate the effect of the common mode currents, we developed a distance averaging method, while, for characterizing the antenna, we used a single-antenna method. For the experimental validation of the system in real conditions, a continuous monitoring of the lake was carried out. During the monitoring, multiple events generated by incursions of motor boats were successfully detected and recorded.

This paper presents a detailed experimental study on operability of a target tracking system based on YOLOv4 for target detection. This work aimed to determine multi-dimensional regions where the two parameters of YOLOv4 algorithm, detection threshold and IOU threshold, allow obtaining the maximum of performance from the detection algorithm, represented by 100% detection for the overall possible detected targets. Low values of these parameters allow more detections, but also increased the number of misclassifications, counted as wrong detections. On contrast, high values of these parameters increase the detection accuracy, but on the cost of total number of detected targets. Considering that each of these parameters can have at least 9 values, 0.1 to 0.9, resulting in 81 combinations, the identification of local and global maxima and minima and of an operational zone, in which the system performance is stable, could be considered as priorities on context of optimal use of the overall target tracking system. Given the high diversity of graphics representation for the system performance in respect with the variance of these two parameters, each of them influencing in an inter-dependent mode the overall performance of the system, the authors propose that the results of each experimental complex scenario to be consolidated in 3D graph, named decision space, one for each environmental context, in which each point is the result of system efficiency reported to a specific pair of values of the mentioned algorithm parameters. These decision spaces present in aggregated form the performance obtained in a target tracking scenario, related on a specific environment, but considered all possible parameter values. The results can be used as a method of pre calibrating the target tracking system prior of effectively using it for its main purpose.

No abstract