Building Ubiquitous Computing Environment in Downtown Improvised Sensor Networks Nobuhiko Nishio#1),#2) #1)Japan Science and Technology Agency PRESTO 21: Precursory Research for Embryonic Science and Technology #2)Ritsumeikan University, Dept. of Computer Science 1-1-1 Noji-Higashi, Kusatsu-shi, Shiga 525-8577, Japan Tel. +81-77-566-1111 (Ext. 7466), Fax. +81-77-561-2669
[email protected]
Abstract In this paper we are proposing to build a ubiquitous computing environment using downtown improvised sensor networks. Our approach is three-fold, we are researching protocols in order to cooperate processing between different nodes in this dynamic environment. We are also developing applications, such as navigation support, as well as an n-1 terminal network simulation testbed for ad-hoc wireless sensor nodes.
1. Introduction Due to the rapid research developments in sensor networks the latest years, the possibilities for various applications in ad hoc networks have been expanded. We are aiming towards applying sensor network technology in order to realize a ubiquitous computing environment in the streets. The cities are saturated with people carrying mobile terminals such as mobile phones, and PDAs. This is in particular the case in the major cities of Japan, and in many areas the potential environment for creating ad hoc sensor networks is therefore already in place. Furthermore, we are also engaged in designing a network connecting vending machines spread out through the cities with optical fibers. These vending machines are also well represented along the streets in Japan, and are therefore very promising candidates for
functioning not only as sink nodes for sensor networks, but also as landmarks for locationbased services for the mobile nodes. We are considering adaptive advertisements, personal navigation and outdoor game entertainment applications as examples of possible applications in this environment. However, it is a challenging task to run applications over an ad hoc sensor network platform where all nodes are moving around without any relationship to the context of the application. It is not likely that the movement of any network nodes can be controlled in order to support a specific application, and the system must therefore be able to adapt to the relentless movement of the terminals. Below we will describe three research topics that we are focusing on for this project, (1) A protocol for dynamic clustering in this environment, (2) Potential applications, (3) An N-1 network simulator we are developing for evaluation purposes.
2. Protocols In this improvised downtown environment, the movement of the individual nodes has no relationship with the interests of the applications. For instance receiver candidates for advertisements are not necessarily located nearby the store that would like to distribute advertisements, there might not even be a constant path through the network between
these nodes. For these kinds of situations, we are constructing the hot-zone clustering protocol where the application sets up a dynamic cluster in the area of interest, and where it would like to target nodes. As long as the hot-zone exists, this cluster is maintained through a cooperative protocol between the cluster-center nodes that are located inside the hot-zone cluster, and the cluster-frontier nodes that are located in the nearby vicinity. In particular, even if all cluster-centers disappear, the cluster-frontiers persistently waits for an opportunity to reconstruct the cluster by increasing the number of cluster-frontiers, and extending the time-out period before the cluster is abandoned.
3. Sample Application As a specific application, we are constructing a personal navigation system. Assuming vending machines in the role as sink nodes and landmarks, a navigation path is set up among N vending machines, where the mobile nodes located between vending machine i and vending machine i+1 cooperate in leading the user node to the destination. The user node requests neighbor mobile nodes to create a route towards vending machine i+1, and the user node is then lead towards the node with the shortest number of hops by monitoring the fluctuations in signal strength of this node. If there are several nodes with the same number of hops, the application leads the user node so that it does not move away from all the nodes. At the same time, vending machine i+1 creates a hot-zone cluster around itself, and sends out supportive information to help the user node in its approach.
4. Evaluation Tools To evaluate our system in such a wireless network composed of several mobile nodes, network simulators are generally used. This is
because an implementation of an environment using real nodes, are subject to the restraints of all the nodes moving, and this is difficult both from the perspective of the number of nodes and also from the perspective of space. In order for this evaluation environment to approach a real node environment as much as possible, we are constructing a hybrid environment with one real node and N-1 nodes simulated inside the network simulator. In this mixed-reality approach, virtual packets that are transmitted inside the simulator are actually transmitted from the simulator host to the real node if it is located in range of these sending nodes. The simulator host receives real packets transmitted from the real node, and transmits these packets as virtual packets into the simulator. Our implementation is based on ns-2 emulation, but we are adding wireless network extensions, mobility support interacting with a real world and simulation of the received transmission strength.
References [1] G. Lambertsen, N. Nishio, “Hotspot Cluster Set-up and Preservation in an Ad-hoc Mobile Sensor Network with a High Degree of Unintentional Movement”, OS-93, IPSJ, May 2003 [2] N. Nishio, “Ubiquitous Computing in the Streets Supported by Network of Vending Machines”, Workshop on Location Aware Computing, UbiComp 2003. [3] N.Nishio, E. Takimoto, and G. Lambertsen, “N-1 Network Simulation: A Mixed-Reality Approach for Wireless Ad-hoc Network Performance Evaluation “, demo at MobiCom 2003, September 2003. [4] ns-2 network simulator, http://www.isi.edu/nsnam/ns/
