Architecture

Overview

The architecture of this sensor network is described in Figure 1. Every sink is connected to the Internet network, which allows them to transmit the sensor information to a centralized database server. For this purpose, several different ways to do it are possible. It slightly depends on the geographical location of the sink and the physical available communication infrastructure. Four standard technologies have been used: GPRS access, ADSL access, Analogue Modem access or direct connection to the Internet. For all these solutions, the TCP/IP stack has to be implemented. Consequently, an embedded PC implementing the TCP/IP stack has to be part of the sink. Also, the data will have to be buffered inside this computer in case of failure or disconnection of the TCP/IP network. For this reason, a database (Mysql) has been used. Finally, the port RS232 will interface between the embedded computer and the node. The programming language which has been used is Java, principally because of its portability. Furthermore a large number of APIs are available, which reduces the development time. Globally, the sinks collect the data coming from the ad-hoc network and store them into local databases. Periodically, data are sent to the centralized database server through the TCP/IP stack. During the whole process the integrity of data is guaranteed.

Detailed architecture of the software

Data coming from the sensor are routed to the sink node via the multi-hop routing protocol. Then, the sink node will transit the messages through the RS-232 interface to an embedded computer called sink computer (the sink node and the sink computer are geographically at the same location). The sink computer has the following main logical functionality: MySQL database, TCP/IP stack and a Java virtual machine. The MySQL database is essential because the TCP/IP connection is not 100% reliable. Consequently, data should be saved locally on the sink computer and should be sent as soon as the TCP/IP connection is restored. The database server collects the information coming from the multiple sinks. It is important to note that it is always the designated sinks that will set up the connection to the server. This is because the sinks will probably be hidden behind a Network Address Translation (NAT) server which shares a single IP address to many end users. Then, the General Manager is the tool used to control and manage the totality of the architecture. This application will be a kind of log centre. Every log message generated by the sink application will be automatically sent to it. Another functionality of General Manager is to control the Wireless Sensor Network where many parameters may be changed. And finally, the server database can be accessed easily and user-friendly from anywhere on the Internet.