ORBIT: A Platform for Smartphone-Based Data-Intensive Sensing Applications

Abstract

Owing to the rich processing, multi-modal sensing, and versatile networking capabilities, smartphones are increasingly used to build data-intensive embedded sensing applications. However, various challenges must be systematically addressed before smartphones can be used as a generic embedded sensing platform, including high power consumption, lack of real-time functionality and user-friendly embedded programming support. This paper presents ORBIT, a smartphone-based platform for data-intensive embedded sensing applications. ORBIT features a tiered architecture, in which a smartphone can interface to an energy-efficient peripheral board and/or a cloud service. ORBIT as a platform addresses the shortcomings of current smartphones while utilizing their strengths. ORBIT provides a profile-based task partitioning that allows it to intelligently dispatch the processing tasks among the tiers to minimize the system power consumption. ORBIT also provides a data processing library that includes two mechanisms namely adaptive delay-quality trade-off and data partitioning via multi-threading to optimize resource usage. Moreover, ORBIT supplies an annotation-based programming API for developers that significantly simplifies the application development and provides programming flexibility. Extensive microbenchmark evaluation and three case studies including seismic sensing, visual tracking using an ORBIT robot, and multi-camera 3D reconstruction, validate the generic design of ORBIT.

Multi-tenant mobile offloading systems for real-time computer vision applications

“…A mobile offloading system provides the capability to process workloads remotely on servers. According to how an application is partitioned across the mobile device and the server, mobile offloading frameworks can be classified into two classes: coarse-grained [56,137,60] and fine-grained offloading frameworks [112,41,55,83,109,121].…”

“…Dynamic profiling and partitioning techniques perform better in such highly varying conditions. It has been adopted in many frameworks that adapt to different goals and constraints [41,39,109,121].…”

“…For example, running an image classification on board using AlexNet consumes around twice the energy consumed when offloading the task using WiFi [82]. 4 Power consumption can be further reduced by profiling the computation and communication overheads, and making dynamic offloading decisions accordingly [112,41,109].…”

Kestrel: Video Analytics for Augmented Multi-Camera Vehicle Tracking

“…To reduce the size of the FC layer, we reduce the number of filters and the number of outputs in the network configuration and re-train. To reduce the memory footprint of the FC layer, we can also split the computation in some layers [29]. Specifically, in the FC layer, the input vector of previous layer is multiplied by the weight matrix to obtain the output vector.…”

DXTK: Enabling Resource-efficient Deep Learning on Mobile and Embedded Devices with the DeepX Toolkit

“…Approaches include the development of sensing algorithm optimizations such as short-circuiting sequences of processing or identifying efficient sampling rates and duty cycles for sensors and algorithm components like [29,40,41]. Other works such as [39] have looked at combining such optimizations with careful usage of energy efficient hardware.…”