As a result of constant demand for the increasing amount of information, traditional centralized computing architectures suffer from lack of robustness, duplication of information, and inadequate system scalability and flexibility. Distributed Intelligent Systems (DIS) show potential in developing management of smart buildings, intelligent transportation, and many other application domains. Peer-to-Peer (P2P) information sharing system, a distributed system that is based on collaboration of peers and efficient use of local computing resources, has the potential to elegantly solve performance issues associated with centralized architectures. In this thesis, we described a prototype based on the Kademlia P2P protocol and a geographic binary tree to illustrate the potential of P2P networks in a distributed intelligent system for control. We implemented a six-node embedded test bed and used it to evaluate the information publishing/querying functionalities. Three sensing peripherals were utilized to demonstrate the flexibility and robustness of the system. These contributions were combined to demonstrate that P2P networks offer a valuable technology for the development of distributed intelligent systems.
展开▼
