With the dissemination of inexpensive sensors and portable computers, such as Personal Digital Assistance (PDA) and laptop PC, we have faced with ubiquitous networks in which every node in environments has its own information and capability to communicate with other nodes via wireless connections. Furthermore, such nodes can move around if a human takes along a portable computer or an animal is attached with a sensor. The information on the nodes seems to change frequently in accordance with their own conditions and surrounding environments. If we can gather these information from a large number of mobile nodes in the environment, we can accomplish a system that monitors the environment or navigates nodes depending on their conditions. However, there is no scheme suitable for periodically and uniformly gathering information from mobile nodes. Although ALOHA and CSMA/CA are famous access control schemes for wireless communications, their performances degrade with the increase of node mobility and the number of nodes. In this paper, we propose a novel scheme to efficiently gather information from mobile nodes. To reduce the impact of data loss incurred by collisions, a terminal, called a reader, first gathers identification (ID) from a node and registers it to its own ID list when the node enters its access area. This is a probabilistic approach where the reader controls response probability that a node sends its ID to avoid collision. Then, it selectively collects data from the node by designating with the corresponding ID. By dynamically switching these two phases, the reader periodically and uniformly gathers data from a large number of mobile nodes one by one. It removes the ID if it detects the departure of the corresponding node. Thus, the performance of the propose scheme highly depends on the accuracy of the ID list. We first analyze the number of nodes that are in the area but their IDs have not registered, called non-registered nodes, and the number of nodes that left the area but their IDs are still registered, called non-deleted nodes. It is originally difficult to derive these parameters because it requires to solve complex state equations. To tackle this problem, we use inversion formula of Palm calculus. We conduct simulation experiments to verify the analysis and compare the proposed scheme with the existing schemes. The throughput of slotted ALOHA becomes 37 % while that of the proposed scheme becomes 74 %. Further, simulation results show that the proposed scheme performs well in a wide range of mobility by appropriately controlling the response probability.

Yusuke Hirano, A Scheme for Continuous Data Gathering from Mobile Nodes, Master's Thesis, March 2007.