FoggyCache: Cross-Device Approximate Computation Reuse

Abstract

Mobile and IoT scenarios increasingly involve interactive and computation intensive contextual recognition. Existing optimizations typically resort to computation offloading or simplified on-device processing. Instead, we observe that the same application is often invoked on multiple devices in close proximity. Moreover, the application instances often process<i>similar</i> contextual data that map to the<i>same</i> outcome. In this paper, we propose<i>cross-device approximate computation reuse,</i> which minimizes redundant computation by harnessing the "equivalence'' between different input values and reusing previously computed outputs with high confidence. We devise adaptive locality sensitive hashing (A-LSH) and homogenized <i>k</i> nearest neighbors (H-kNN). The former achieves scalable and constant lookup, while the latter provides high-quality reuse and tunable accuracy guarantee. We further incorporate approximate reuse as a service, called \name, in the computation offloading runtime. Extensive evaluation shows that, when given 95% accuracy target, \name\ consistently harnesses over 90% of reuse opportunities, which translates to reduced computation latency and energy consumption by a factor of 3 to 10.