Structural Monitoring using Wireless Sensors and Controller Area Network Gregory Jaman SMT Research 200-135 Innovation Dr. Winnipeg, Manitoba, R3T 6A8 jaman@ee.umanitoba.ca Sajid Hussain Computer Science Acadia University Wolfville, Nova Scotia, B4P 2R6 Sajid.Hussain@acadiau.ca Abstract— This paper investigates implementation and design issues for a heterogeneous network for structural monitoring. The proposed application uses wireless sensors and the controller area network (CAN) to provide energy efficient monitoring. The contributions include: implementation of the PIC18f4680 platform in TinyOS, creating a message protocol over CAN, and developing a sensor board. I. I NTRODUCTION Structural health monitoring is an on going topic for re- search in wireless sensor networks. In order to get accurate data about a structure, the sensors typically need to be em- bedded within the structure, in locations that are not easily accessed without demolition. Because of these constraints, sensing lifetime of the network must be reasonable in compar- ison to building life. It would not be unreasonable for building developers to demand a sensing lifetime in the order of 10’s of years. Wired Sensor Networks in the past have been inexpensive and more reliable compared to Wireless Sensor Networks (WSNs). However, in recent years, WSNs are continually becoming less expensive and more dependable. As well, advances in signal-processing techniques and new network architectures are making WSNs more reliable. In contrast, the cost of wired networks has remained constant or increased as there is an increase in the cost of the wire, the connectors, and the labor to connect point A to point B. WSN’s lifetime is limited by the battery and power consumption whereas a wired network has potentially infinite lifetime. Coverage and reliability is also limited in WSN. Although techniques such as multi-hop routing can be deployed to increase coverage, it comes at the cost of the battery power. A heterogeneous network can take advantage of the benefits of each network [1]. A term, Civionics, has been recently introduced to de- scribe the discipline of monitoring structures. Civionics is an emerging discipline describing the interaction between civil engineering and electronics. Integration of intelligent sensing will allow civil structural engineers to evaluate new concepts, materials, and understand aging buildings. The demand for this technology is evident in problems such as moisture intrusion in sealed building envelope systems. New buildings being constructed with efficiency in mind are sealed such that air does not easily pass through walls. Any intrusion of water is absorbed into the walls and not allowed to evaporate, resulting in the wood frame rotting or mold growing. The effects of moisture in building structures are out of the scope of this thesis, but is the motivation behind the sensor network and its application. A method for integrating wireless nodes in an existing wired sensing network for detecting moisture in building structures is proposed in this paper. The benefits of wireless nodes can be seen in installation cost and maintenance. A typical installation of a wired network for moisture detection in a building in shown in Figure 1. Segments of Moisture Detection Tape (MDT) are installed along the perimeter of the building. These segments of tape act as sensors providing data to the Remote Measurement Units (RMU). Significant time and cost is required to route wires in stud cavities to sensors, between RMU’s, and back to the gateway (Detec 6000). Installation costs have been valued at approximately $50USD per segment of tape, overshadowing the cost of introducing wireless components to the hardware. In addition, because of the nature of the measurements, electrical isolation is extremely important. Resistive values measured on the tape range from 100KΩto 100MΩ. By sharing common power and communication wiring between nodes, accidental grounding to building ground introduce errors over this large measurement range. Accidental grounding is typical in large commercial structures. Tradesmen not familiar with the electronics or unknowing can compromise the installation by faulty wiring or grounding the tape. Depending on the size of structure, it is not uncommon to have over 1000 segments. It should be evident how accidental wiring errors can easily occur. Implementing a pure wireless network is not always op- timal. In certain instances harsh environments can limit the communication range or prohibit devices from operating in the same frequency band. While in some instances wireless communication may fail when loss rates become unacceptable. To span large distances multi-hop routing is required in a mesh network topology. Multi-hop routing requires complex algorithms to synchronize communication and increased power consumption due to listening for neighbor’s requests. In these cases an option for a node to communicate over a wired medium is necessary.