Adaptive medical feature extraction for resource constrained distributed embedded systems

Tiny embedded systems have not been an ideal outfit for high performance computing due to their constrained resources. Limitations in processing power, battery life, communication bandwidth and memory constrain the applicability of existing complex medical/biological analysis algorithms to such platforms. Electrocardiogram (ECG) analysis resembles such algorithm. In this paper, we address the issue of partitioning an ECG analysis algorithm while the wireless communication power consumption is minimized. Considering the orientation of the ECG leads, we devise a technique to perform preprocessing and pattern recognition locally on small embedded systems attached to the leads. The features detected in pattern recognition phase are considered for classification. Ideally, if the features detected for each heart beat reside in a single processing node, the transmission will be unnecessary. Otherwise, to perform classification, the features must be gathered on a local node and thus, the communication is inevitable. We perform such feature grouping by modeling the problem with a hypergraph and applying partitioning schemes. This yields a significant power saving in wireless communication. Furthermore, we utilize dynamic reconfiguration by software module migration. This technique with respect to partitioning enhances the overall power saving in such systems. Moreover, it adaptively alters the system configuration in various environments and on different patients. We evaluate the effectiveness of our proposed techniques on MIT/BIH benchmarks.