THE scaling of device technologies has made possible significant increases in the embedding of computing devices in our surroundings. Embedded microcontrollers have for many years surpassed microprocessors in the number of devices manufactured. The new trend, however, is the networking of these devices and their ubiquity not only in traditional embedded applications such as control systems, but in items of everyday use, such as clothing, and in living environments. A trend deserving particular attention is that in which large numbers of simple, cheap processing elements are embedded in environments. These environments may cover large spatial extents, as is typically the case in networks of sensors, or may be deployed in more localized constructions, as in the case of electronic textiles. These differing spatial distributions also result in different properties of the networks constituted, such as the necessity to use wireless communication in the case of sensor networks and the feasibility of utilizing cheaper wired communications in the case of electronic textiles. Electronic textiles, or e-textiles, are a new emerging inter disciplinary field of research, bringing together specialists in information technology, microsystems, materials, and textiles. The focus of this new area is on developing the enabling technologies and fabrication techniques for the economical manufacture of large-area, flexible, conformable information systems that are expected to have unique applications for both the consumer electronics and aerospace/military industries. They are naturally of particular interest in wearable computing, where they provide lightweight, flexible computing resources that that are easily integrated or shaped into clothing. Due to their unique requirements, e-textiles pose new challenges to hardware designers and system developers, cutting across the systems, device, and technology levels of abstraction:

The need for a new model of computation intended to support widely distributed applications, with highly unreliable behavior, but with stringent constraints on the longevity of the system.
Reconfigurability and adaptability with low computational over head. E-textiles must rely on simple computing elements embedded into a fabric or directly into active yarns. As operating conditions change (environmental, battery lifetime, etc.), the system has to adapt and reconfigure on-the-fly to achieve better functionality.

Device and technology challenges imposed by embedding simple computational elements into fabrics, by building yarns with computational capabilities, or by the need for unconventional power sources and their manufacturing in filament form.

In contrast to traditional wearable computers, which are often a single monolithic computer or a small computer system that can be worn, e-textiles will be cheap, general purpose computing substrates in the form of a woven fabric that can be used to build useful computing and sensing systems "by the yard" [1].

Techniques to program such networks are required that permit useful applications to be constructed over the defect and fault-prone substrate. There is a need for a new model of computation to support distributed application execution with highly unreliable behavior at the device-level, but with stringent constraints on longevity at the system level. Such a model should be able to support local computation and inexpensive communication among computational elements.

In the classical design cycle (Fig. 1), the application is mapped onto a given platform architecture, underspecified constraints (performance, area, power consumption).When these constraints are met, the prototype is tested, manufactured, and used for running the application.