In this paper, a clustering algorithm called the Reconfigurable Dominant Sequence Clustering (ReDSC), is proposed. The experiments show that ReDSC reduces the parallel run time efficiently by 30 percent. Moreover, an empirical comparison of ReDSC with its peer in the multi-processor model, i.e., DSC algorithm, is provided and the results demonstrate the 35 percent improvement of ReDSC over DSC. Then, a hardware implementation of a dynamic scheduler for run-time communication-aware scheduling of hardware tasks is proposed. The proposed scheduler makes use of the ReDSC algorithm for allocating tasks to the processing reconfigurable cores, dynamically considering the communication costs. The simulation results illustrate the benefits of the proposed scheduler as compared to that of other static and basic schedulers in the technical literature.

Sensor Networks have great potential for many applications. In time-critical and security applications, it is important to have on time delivery, minimum access time and maximum data rate. Most of sensor networks have a large number of sensor nodes which considering the memory and processor of each of them all together they provide a huge data storage space and also high processing capability. In this paper we propose a method to implement the parallel data monitoring and processing on a large number of sensor nodes in a sensor network that can be used is a security environment. We can use the sensor, memory and processing elements of each sensor node store and process a part of large database or computation. The data which is entered to the database can come from the sensors of sensor nodes. By this method not only we can use the memory of each sensor node to store apart of our data but also we can do parallel processing on the stored data of all sensor nodes. The results of our simulations show this method provides a huge storage space to store a considerable amount of data and also speed up the computations.