Abstract:This paper investigates an integrated sensing-communication-computation (ISCC) based multi-view-multi-task (MVMT) edge AI inference system. Each device senses a narrow view of a target area and processes the echo signal to generate real-time sensory data. An edge server receives and combines multiple views of data from multiple devices to complete several downstream inference tasks. Compared with existing designs where dedicated sensory data is obtained, transmitted, and processed for each task, this ISCC-based MVMT framework enjoys reduced costs of sensing, on-device computation, and communication overhead due to data sharing among different tasks. The challenges of improving all tasks’ inference accuracy lie in the tight coupling of sensing, communication, and computation among different devices and sensory view competition among different tasks. These two challenges intertwine, making the multi-task optimization problem mixed-integer non-convex programming. To tackle this problem, we propose a joint device scheduling and resource allocation (JDSRA) scheme, which alternatively solves a subproblem of joint device scheduling and time allocation and a subproblem of resource allocation till convergence. Particularly, in addition to a dynamic-programming-based optimal device scheduling algorithm, a low-complexity suboptimal algorithm is proposed based on sorting a derived closed-form indicator, which represents the increase of all tasks’ inference accuracy per time unit consumption. Besides, a low-complexity optimal resource allocation algorithm is proposed by parallelly solving multiple simple convex subproblems. Numerical results based on jointly completing three tasks of human motion recognition, human height recognition, and localization in smart home scenarios are conducted to verify the performance of our proposed schemes.

Adaptive joint configuration optimization for collaborative inference in edge-cloud systems

Extendable Multi-Device Collaborative Pipeline Parallel Inference in the Edge-Cloud Scenario

AI-assisted Action in Edge Computing System: A Joint Latency and Accuracy Oriented Approach

Joint Model and Data Adaptation for Cloud Inference Serving

Joint Communication and Computation Resource Allocation for Cloud-Edge Collaborative System.

Collaborative Cloud and Edge Computing for Latency Minimization

JAVP: Joint-Aware Video Processing with Edge-Cloud Collaboration for DNN Inference

Joint Optimization With DNN Partitioning and Resource Allocation in Mobile Edge Computing

Joint Device Scheduling and Resource Allocation for ISCC-Based Multi-View-Multi-Task Inference

Multi-Compression Scale DNN Inference Acceleration based on Cloud-Edge-End Collaboration

A collaborative cloud-edge computing framework in distributed neural network

Collaborative Inference for Deep Neural Networks in Edge Environments

An Adaptive Neural Architecture Search Design for Collaborative Edge-Cloud Computing

Adaptive Data Sharing and Computation Offloading in Cloud-Edge Computing with Resource Constraints.

Joint Configuration Adaptation and Bandwidth Allocation for Edge-based Real-time Video Analytics

Joint optimization of multi-dimensional resource allocation and task offloading for QoE enhancement in Cloud-Edge-End collaboration

Dependent task offloading mechanism for cloud–edge-device collaboration

Adaptive Configuration Selection and Bandwidth Allocation for Edge-Based Video Analytics

Task allocation algorithm and optimization model on edge collaboration

Joint Task Offloading and Resource Allocation for Quality-Aware Edge-Assisted Machine Learning Task Inference

Joint DNN Partition and Resource Allocation Optimization for Energy-Constrained Hierarchical Edge-Cloud Systems