Dissertation Talk: Adapting Swarm Applications

Seminar: Dissertation Talk: EE | May 8 | 3-4 p.m. | 531 Cory Hall

 Ben Zhang, UC Berkeley

 Electrical Engineering and Computer Sciences (EECS)

The swarm refers to the vast collection of networked sensors and actuators installed in our connected world. Many swarm applications transport, distill, and process large streams of data across the wide area in real time. The increasing volume of data generated at the edge challenges the existing approaches of directly connecting devices to the cloud.

In this talk, I will first present architecture design trends and challenges for developing swarm applications. Specifically, I will focus on the challenges from the scarce and variable WAN bandwidth and the heterogeneous compute environment (from low-power micro-controllers to powerful compute units). When network resources or compute resources are insufficient, non-adaptive applications will suffer from increased latency or degraded accuracy. Existing approaches that supports adaptation require extensive developer efforts to write manual policies or are limited to application-specific solutions.

We present a systematic and quantitative approach to build adaptive swarm applications. Our solution includes three stages: (1) a set of programming abstractions that allow developers to express adaptation options; (2) a combination of offline and online profiling tool that learns an accurate profile to characterize resource demands and application utilities; (3) a runtime system responsive to environment changes, maximizing application utility based on the learned profile. We evaluate the effectiveness of adaptation with several swarm applications: pedestrian detection, augmented reality, and monitoring log analysis. Our experiments show that all applications can achieve low latency responses with nominal accuracy drops.

 CA, benzh@cs.berkeley.edu, 510-5175530