Energy efficiency in network communication is critical for wirelessly connected small computing devices, which run on limited battery capacity. Under realistic movement scenarios (e.g., a person traveling at airplane, automobile, or biking speed), a mobile sender can track its own movement and postpone communication (subject to application deadline constraints) until it moves close to the communication target. This will save significant energy of sending, which grows superlinearly with the communication distance in, say, the single hop wireless context. However, movement tracking requires the mobile device to be turned on and hence consumes energy. Instead of continuous tracking, the mobile device should sample its movement and be allowed to sleep between the sampling instants (provided that the application also does not have work to do during the sleep). In this paper, we present an adaptive scheduler for determining an effective sampling schedule given changing operating conditions. Our experimental results show that the scheduler can achieve substantial energy savings over a device that is always on. Moreover, the scheduler?s adaptivity allows it to outperform fixed sleep periods between tracking, since the "right" sleep period depends on dynamic system conditions and cannot be determined a priori.