Constraints and Challenges

Unlike traditional computing devices, e-textile presents a different set of challenges and constraints.

Fault Tolerance
E-textile, due to its nature, is inherently a distributed embedded system with multiple computing nodes interconnected by communication channels. However, because e-textiles are manufactured in a low cost and large-scale manner, it is susceptible to manufacturing defects. These defects could be in the computational and the sensor nodes or the interconnecting channels that link them. Furthermore, normal wear and tear during usage can easily damage the electronics.

The challenge is therefore to be able to carry out the applications even when some of the nodes and interconnects fail. Because of the low cost of production and mass “embedding” of processors and sensor nodes, there is sufficient redundancy in the e-textile. One possibility is to get the nodes to reconfigure and reroute its communication channels and repartition its computational load.

Energy Consumption
Power consumption has to be kept low, so that there is no need to carry a heavy battery pack and additional cooling mechanisms. This is particularly important when it involves a human being wearing an e-textile suit.

There are also innovative methods of generating power, e.g. tapping a human’s kinetic energy or using thermal conduction. These methods are still at a stage where the power output is limited. To incorporate them, e-textile has to strive for minimum power consumption.

The power consumption can be managed at several levels. Other than processors’ micro-architecture, the processing requirements of the software and the communication needs of the network all play an important role in minimizing power consumption. The challenge is find a system architecture that can do that.

Processing Capability
Cost, size and power constraints limit processing power available on a single node. While miniaturization continues to enhance the capability of the computational elements on the e-textile, parallel processing is still required to maximize the computational power by utilizing all the available nodes. Coming from the energy perspective, overall computation has to be kept to the minimum to conserve power.

The challenge is therefore how to derive the maximum required computational power to carry out meaningful applications, while keeping energy consumption low. This requires power or energy simulations to find the optimal point where both constraints can be satisfied. In a step further, a range of optimal points may be needed to cater for changing computational loads and changing energy supply.

Inter-node Communications
Like processing power, communication bandwidth is a limited commodity. Inter-node communication has to be kept to the minimum in order to conserve power. But in order to carry out parallel processing, some communication overhead has to be incurred. In addition, reliability issues may impair communications among nodes.

The challenge is therefore to find a topology and protocol that can handle faults in the communication links, as well as reconfigure itself to deliver information across the textile in a way that minimizes power consumption.