We are attempting to address the computing research issues that arise in the successful realization of e-textiles. Our approach to this process is to build physical prototypes and employ extensive simulation to explore the vast design space in e-textiles. Towards that end, we are constructing a modeling and simulation environment to explore that space. Specifically, we are interested in issues such as:
- Can we identify a computing architecture(s) appropriate for e-textiles? How do we arrive at and justify such an architecture?
- Can we exploit the nature of the textile manufacturing process to build low-power, reliable communication networks in fabric?
- At what point, if any, does it make sense to use Bluetooth-style radios to augment the wired communication scheme in e-textiles?
- What software services are necessary for most e-textile applications and how can they be supported on a range of fabrics?
- How can we best explore the design space for e-textiles? How can we modify and employ existing modeling and simulation tools towards that end?
We are designing and building several prototype e-textile applications including:
- A large-scale acoustic array for sound localization/enhancement,
- An ultrasonic emitter/detector array for location/motion sensing,
- A glove keyboard, and
- An omidirectional videosystem.
We envision two main types of applications of e-textiles, wearable computing and large-scale sensor networks. E-textiles-based wearable computers will sense the user's motion and be aware of the objects around the user, while e-textiles-based sensor networks must be reliable in harsh environments and be extremely low power to be able to last days and weeks in the field without maintenance. Both types of applications represent an extreme corner of the distributed computing design space: physically spread over a relatively smaller space, but having a greater dependence on physical locality of computation, lower bandwidth for communication, and less available energy. The stricter constraints faced by e-textiles require new solutions to be found for questions that have been studied in embedded systems and distributed computing. They also create problems that have not been studied before, such as optimizing energy usage when both the power sources and power consumers are distributed throughout the system, or allocating tasks to processing and sensing elements located on the body based upon the motion of a user and objects in the user’s environment.