1.1 Introduction

A team of kineticists from the same laboratory work to study of the effects of daily walking exercise among workers with sedentary jobs. The researchers hope to monitor changes in metabolic rate as the participants go from almost no daily exercise to between thirty and forty-five minutes of walking each day. Each of the participants dons a chest strap, a slightly hefty belt-pack, and an ear clip before going for their daily walk. The equipment in the fanny pack records the data over the sixty-day trial period for later analysis by the scientists.

Meanwhile, conservationists in Rwanda work to save the mountain gorillas living in Virunga National Park. Refugees from the surrounding war-torn areas continue to poach the gorillas, threatening the survival of the species. The park officials have created a sensor network that includes several digital cameras, solar battery chargers, and communication equipment. In well-hidden locations throughout the park, each station monitors several square miles of forest. At night, the infrared abilities of the cameras allow for the continuing protection of the gentle beasts. The data are relayed every hour to park headquarters or whenever the system detects that a poaching may be in progress. The quick reporting of these situations allows the Park authorities and environmental groups to intervene before a slaughter occurs or apprehend the criminals, preventing further damage by the same group.

The three systems presented in the imagined scenarios above differ considerably. The Birkebeiner configuration requires equipment and packaging that can withstand the cold, wet conditions present when mounted close to the snow as well as motion and shock challenges. The relatively constant data streams require high-bandwidth connections. The walking study requires low maintenance and low power equipment. It may not demand as much bandwidth as the skiing experiment, but the connection must be reliable to make the study worthwhile. The African setup must run for an extended time without attendance and ideally be self-powered. It must perform image analysis of sufficient complexity to determine that it should alert authorities to a potential poaching.

Despite their significant differences, the three sensor networks described in the previous section perform the same fundamental functions. Data are collected from a network of diverse sensors, stored, in some cases analyzed locally, and finally transmitted to a system capable of more powerful computation, for monitoring and study. The researchers depicted above are not interested in building sensor networks; they are studying elite athletes, improving quality of life, and saving a disappearing species. The sensors each group must use often exist, but as an inharmonious jumble of devices. Often, connecting them requires a Herculean engineering effort. Outsourcing the construction is often too expensive given the short length or small budget of the study. The Snap! toolkit takes advantage of the commonality among the systems to reduce the effort required to network the devices.

The Snap! toolkit is designed to make constructing small, human-scale sensor networks fast and easy. Snap! standardizes and specifies the common functions of nodes on sensor networks, such as reporting data and announcing presence. Snap! permits enough flexibility at the node level to allow almost any type of sensor to be attached to the network. It also describes the essential characteristics of the communications link between the nodes. A shelf of components built following the Snap! specification would allow the creation of a customized sensor-network in a matter of days or weeks, instead of several months. Snap! is not the ideal solution for all sensing applications and does not try to be. Emphasis has been placed on hardware and software reuse and rapidity of design, not suitability for particular size, power, or bandwidth requirements. For instance, Snap! would not be an appropriate solution for developing a "sensor shirt" intended for constant wearing as a marketed product. However, during development of the shirt Snap! would be a valuable asset in determining an appropriate suite of sensors and final networking setup. Using a Snap! platform, the data gathering problems mentioned above can be plausibly addressed.

Similarly, customer feedback also influences the designs of snowboard manufacturers. From the effectiveness of a new binding system to the effects of a deeper sidecut on turning control, conditioned riders report their opinions on the effects of a change in design. However, when attempting to monitor the equipment objectively, the weight and bulk of the sensing and telemetry systems prevented the testers from using the equipment naturally, resulting in invalid testing data (P. Maravetz, Burton Snowboards, Sports Product Development Lecture, January 18, 2000).

Product design, from everyday apparel to the most advanced athletic gear, could benefit from monitoring variables that are relevant in the particular design. While the technology exists for all developers to design custom systems, few companies actually monitor their equipment in the field. They take most of their measurements in a lab. A sensor toolkit would allow designers and testers to choose among existing sensors needed for a particular product test. A toolkit would allow a designer to incorporate a new, non-standard sensor into the network without modification to the other sensors or to the existing sensor network. The result: products that take into account objective and subjective data that perform to an overall more satisfying level.