NestDNN: Resource-Aware Multi-Tenant On-Device Deep Learning for Continuous Mobile Vision

TL;DR

NestDNN is a resource-aware framework for mobile vision systems that dynamically optimizes deep learning models for better accuracy, performance, and energy efficiency amidst changing runtime conditions.

Contribution

It introduces a novel resource-aware multi-tenant on-device deep learning framework that adaptively balances resource-accuracy trade-offs in real-time.

Findings

Up to 4.2% increase in inference accuracy

2.0x faster video frame processing

1.7x reduction in energy consumption

Abstract

Mobile vision systems such as smartphones, drones, and augmented-reality headsets are revolutionizing our lives. These systems usually run multiple applications concurrently and their available resources at runtime are dynamic due to events such as starting new applications, closing existing applications, and application priority changes. In this paper, we present NestDNN, a framework that takes the dynamics of runtime resources into account to enable resource-aware multi-tenant on-device deep learning for mobile vision systems. NestDNN enables each deep learning model to offer flexible resource-accuracy trade-offs. At runtime, it dynamically selects the optimal resource-accuracy trade-off for each deep learning model to fit the model's resource demand to the system's available runtime resources. In doing so, NestDNN efficiently utilizes the limited resources in mobile vision systems to jointly maximize the performance of all the concurrently running applications. Our experiments show that compared to the resource-agnostic status quo approach, NestDNN achieves as much as 4.2% increase in inference accuracy, 2.0x increase in video frame processing rate and 1.7x reduction on energy consumption.