Advancements in microcontrollers, memory, and radio technology have enabled more sophisticated functionality at the sensor node, and by extension, within sensor networks. Even for ultra low-power mixed-signal processors such as the Texas Instruments MSP430, real-time signal processing for demanding Parkinsonian tremor monitoring is now a reality. Post-processing of wireless sensor network data will remain an important aspect of pervasive health monitoring; however, as technology progresses, more processing will be driven toward the node level. With both data and processing capability locally available, intelligent decisions about routing, caching, and source coding can dramatically reduce power consumption. Ultimately, the premise of application-layer context-aware signal processing is to achieve one or both of the following aims: decrease the amount of data sent or decrease how frequently data is sent. Techniques such as directed diffusion have demonstrated noticeably better energy efficiency over other routing schemes such as omniscient multicast using such principles. For certain real-time medical applications such as anomalous gait event detection, real-time signal processing could yield an significant reduction in the quantity of data to be sent. Additionally, when combined with source coding methods such as adaptive pulse code modulation, such efforts could yield an impressive reduction in the amount of data sent. Reducing the quantity of data would have an even larger impact on the communication system’s power demands, and would thus enable systems to run unattended for perhaps, weeks.