| Summary: | 碩士 === 逢甲大學 === 資訊工程學系 === 104 === With the advent of wireless networks and Internet of Things (IoT), many current applications need the cooperation of sensor networks. Comparing to the traditional sensor networks, the wireless sensor networks (WSNs) are more flexible in many respects. In past decades, there are many applications, such as military surveillance, environmental monitoring, healthcare, and smart appliances [1][2], had been proposed, thus make the wireless sensor networks valuable.
In wireless sensor networks, the sensing data are often propagated to the backend Base Station (BS), by single-hop direct or multi-hop relay transmission approach. By this way, the sensor nodes around the BS induce hot-spots, and thus consume their energies quickly. This scenario will result in network broken, and eventually reduce the network lifetime.
Recently, the network topologies with multiple sinks have been proposed to cope with the hot-spot issue, and thus prolong the network lifetime. Initially, a paradigm allowing the sensing data is delivered to any one BS is proposed. This is termed as anycast. Thereafter, for further improving the transmission reliability, k destinations are expected. This is so-called K-anycast. However, in K-anycast scenarios, how to select the most appropriate k destinations from all BSs to deliver data can be a challenge. In this thesis, we thus propose an energy-efficient K-anycast protocol for multi-sink wireless sensor networks.
In our scheme, we assume the sensor nodes are not equipped with GPS. We achieve our protocol via the following two steps: 1.) Firstly, all nodes in the network create its integrated routing table, by using the directional diffusion packets issued from all sinks. 2.) From this routing table, the node can either quickly and roughly select the k delivering paths by using greedy algorithms, or precisely choose the k shortest paths by using time-consuming permutation methods. Besides, for reducing network energy consumption, the protocol further integrates the packets whose delivering paths are overlapped to transmit.
The simulation results show that: comparing to the past work [5], our proposed K-anycast protocol can achieve about 40% improvement on total path hop counts, in various network sizes, different number of sinks (i.e. k), and the same node’s transmission range. In addition, our proposed protocol also achieves about 40% efficacy on actual path length and 50% reduction on network energy, when it is applied in a larger network size and a shorter node’s transmission radius. So we conclude that the proposed protocol can make more contributions and benefits when it is applied on real-time environments.
|
|---|
