In the past decade, wireless sensor network (WSN) has caught a lot of attentions in many research communities due to its unique challenges which require integrated solutions involving different disciplines. Energy efficiency in WSN, for instance, requires hardware and software components of one single embedded device consuming minimum energy to operate on one hand. On the other hand, an energy effcient yet reliable networking scheme which considers the performance of the network as a whole is particularly essential to solve this issue. Traditional application-specific integrated circuit (ASIC) based sensor node architectures provide high computation capabilities with minimal energy consumption. Existing WSN deployments consist of heterogeneous sensor nodes with different classes of computation and communication capabilities to fulfill the application requirements. However, due to the limitations of ASIC based architectures, changing the hardware capability after the sensor nodes are deployed is difficult if not impossible. Certainly, the hardware reconfiguration can be beneficial when the environment to be monitored changes over time and the flexibility of the system is required. The reconfigurable sensor nodes can dynamically change their features according to the run-time situations to meet the requirements of performance and efficiency. Recent advance of the Field-programmable gate array (FPGA) offers reconfigurable technologies which provide deep sleep mode with ultra-low power consumption and fast wake-up time. These features are extremely important for energy critical WSN applications and enable the possibilities to explore new hardware/software architectures for the WSN. The proposed project is intended for investigating the WSN applications and network architectures from a perspective of the reconfigurable sensor platform. The problem statement is that, given a group of reconfigurable sensor nodes, how we can redesign the WSN network architectures to benefit from the reconfigurable capability. To explore this, we plan to perform a series of experiments to obtain real world measurements of the latest FPGA technologies. Based on the findings, we will examine the state-of-the-art WSN network architectures to explore the potential improvements and appropriate application scenarios.
The investigation will focus on the general networking architecture without too many emphases on specific applications. However, a general monitoring application will be chosen as demonstration. The goal is to identify the scenarios which can benefit from the reconfigurable platforms. Currently several points are considered worthy of further investigation:
- Hierarchical WSN: To provide a scalable network which prolonging the network lifetime, hierarchical networking structure is commonly used to balance the work load and reduce the traffic. LEACH and HEED are two famous approaches to form clusters among the sensor nodes. Cluster heads which are elected by the member sensor nodes should aggregate the data collected by the member sensor nodes. With typical ASIC sensor nodes, a cluster head which has the same computation capability with member sensor nodes will spend more time on data processing which means more energy will be consumed. Potential efficiency improvement can be gained using reconfigurable platforms by reconfiguring the elected cluster head with hardware accelerators required by the data processing which can shorten the processing time and turn the node into sleep mode sooner.
- Situation based WSN: In many scenarios, the requirement of sensor node's features changes according to the operation modes of the WSN. However, a typical ASIC sensor nodes can not change their capabilities and the designer have to compromise the efficiency to cover all the situations. For example, a crowd detection system in a museum can turn into a intruder detection system during the night which has di erent security requirement to the communication. The reconfigurable platform offers the possibilities to have optimized sensor nodes for specific situation.
- Energy harvesting WSN: A WSN equipped with energy harvesting device can be considered as a special case of the situation based WSN. A sensor nodes which can harvest more energy from the environment can perform more complex tasks than other nodes. The recon gurable platform can be con gured with different features based on the harvesting rate of the sensor node.