We are developing a set of very low-cost, wireless, wearable sensors that enable a large group of people (e.g., hundreds or thousands) to participate in an interactive musical performance. The sensors themselves are simple piezoelectric accelerometers that detect extremes of limb motion, upon which they transmit a narrow RF pulse. Although one can distinguish between sensors on the upper and lower body by using a different carrier frequency, and zone the locations of activity roughly through carrier strength, we do not plan to independently ID each performer, but instead to measure and react to the characteristics of ensemble behavior. We have built this system and have developed algorithms that use this data to explore techniques of mapping large-group, real-time musical interaction.