A Compact Wireless Modular Sensor Platform Ari Y. Benbasat and Joseph A. Paradiso To simplify the rapid prototyping and testing of wireless sensor systems, it was decided to design a modular sensor platform. Overall, the goal was to allow the user to treat sensing as a commodity, i.e. allowing an application to trivially incorporate different kinds of measurement. There were three keys to achieving that goal: Encapsulation Knowledge: A single pane of a modular system can represent the best practices in a given field save a large amount of design time allow for easy upgrades. E.g., an RF single pane with a proper HF transceiver and antenna layout Reducing repetition of circuit design: The creation of individual panes containing one or more such circuits can eliminate much of the drudgery of the design process: Most systems involve reuse of known circuit blocks usually with only slight changes. E.g. serial line converters, power regulators and microcontroller support circuitry. Simplifying prototyping: Rather than proceed directly to a final layout, this platform makes it possible to build a prototype to: collect the relevant data provide a valuable proof of concept help detect flaws in the design provide a basis to begin the construction of necessary interface and analysis software. Given the above goals, we consider three key philosophies in the individual board designs: Individual panes should be combinations of circuitry that cannot or should not be separated E.g. Six-axis inertial measurement unit It must be as easy as possible to combine and recombine the panes into new applications As many data/signal/power lines as possible Connectors must be structurally strong Expandability is key to future utility Footprint and height should be reasonable Monopolization of interconnects should be avoided Goals & Philosophy Hardware Instatiation The master board is responsible for the data collection and transmission to the central basestation and is included in every project. 22 MIPS processor with 12-bit ADC 916MHz transceiver running at 115.2kBps The processor pins are broken out to the interboard interconnects This board draws 35mW under normal operation. A six-axis IMU is provided to measure motion. Acceleration via two dual-axis accelerometers Angular velocity via three gyroscopes Four-way static tilt sensor for single-bit acceleration measurement This pane draws 65mW. A matched pair of sonar receiver and transmitter boards provide for distance measurement. Single omnidirectional transmitter Two pickups placed a fixed distance apart This configuration allows both displacement and relative angle to be calculated. The transmit and receive modules each draw approximately 120mW. A board is provided for inputs from a number of different tactile and pressure sensors combinations: Four single-ended force-sensitive resistors Two back-to-back FSR bend sensors Two piezoelectric sensors Loading-mode capacitive proximity sensor This pane draws 65mW. On-board data storage is provided. 1Gbit flash memory chip Controlled by a microcontroller via SPI This board draws 40mW. An ambient sensing board provides a range of methods of detecting audible and visible signals. Narrow and wide cone phototransistors Pyroelectric heat sensor Microphone (and microcontroller for processing) VGA quality cellphone camera This board draws 100mW (including the camera). Sample Applications & Lessons Learned The Wearable Gait Lab, designed by Stacy Bamberg, applied our platform to develop a prototype inexpensive wireless wearable system for the analysis of the motion of feet. The system consisted of: Master board (data collection / transmission) IMU (motion capture) Tactile (connected to insole Sonar boards (ranging) Power regulation board (3.3V, 5V, 12V) The wireless nature of this system allows for real-time feedback to the patient during daily wear which is not possible with a fixed gait lab. This application exploited the modularity of the system to examine the benefits of capacitive and sonar sensing that were not considered in the initial design. We consider lessons learned in two categories: Electrical Timing details become quite important as multiple boards are connected. Long sensor startup times can cause problems for polling-based systems. Digital ouputs measurement times can add up Power supplies and regulation also suffer for heterogeneous devices. More voltage levels mean more regulators, leading to increases in RF noise. Voltage level clashes between devices (eg sensor and ADC) also occur. Mechanical Strength of individual connectors is key in wearable applications. Must consistently provide both mechanical and electrical contact. Parts rated for large numbers of insertion cycles are a necessity. Possible orientations of the boards should not be limited Daughter boards and connectors should not interfere. “Top” boards only if absolutely necessary. This platform is currently being used to prototype Real-time Adaptive Sensor Systems. These techniques should reduce the sensor system's power drain by varying: The sampling rate of the sensors The choice of sensors Analysis of the incoming data such that the amount of data collected by the system is reduced without affecting the amount of useful information collected. This work is being tested using a prototype stack consisting currently of the master and IMU boards. Future work will be based around the ambient board and possibly other boards as well. This application exploits the completeness of each sensing modality as expressed by the stack panes. E.g., the IMU board provides both low and high power (and accuracy) sensors for measuring acceleration.