Abstract
Nowadays, physical therapy has become an indicator of the advancement of a country. Physical therapy helps patients to regain and maintain proper functioning of their bodies. Traditionally, physical therapy has to be performed in the hospitals, assisted and monitored by professional therapists to ensure the safety and effectiveness. With the development of microelectronics, home rehabilitation is now possible with wearable sensors worn on patients to monitor their rehabilitative exercises at home. As most these wearable devices communicate their data with the gateway (or sink) through radio, the blockage of radio signals by the human body is a difficult challenge. From the network perspective, the blockage results in intermittent connections of the wireless body area network (WBAN) formed by the wearable sensors, causing data losses during transmissions. Such a WBAN can be modeled as a delay-tolerant network (DTN), and its data loss problem may be solved using many existing solutions proposed in the literature. On the other hand, when considering the specific application of rehabilitation exercise monitoring, more efficient DTN solutions may be developed. One key observation is that almost all rehabilitation exercises require repetitive movements. The repetitive body moving patterns provide valuable hints on when two sensor nodes will meet, and how one may help the other to relay data to the sink. In this thesis, we exploit this behavior for very efficient transmission scheduling of the wearable sensors. We solve the problem in two parts. First, we collect dynamic neighboring information of the wearable sensors during the exercise. From the collected information, repetitive neighboring patterns are identified, which are then used to schedule when two sensors should communicate to relay data. In the second part of our scheme, the wearable sensors just follow the transmission schedule to send and relay data. Experimental results show that our system operates correctly using the transmission schedule and it performs well in real rehabilitation exercises.