View The Document

Accession Number:

AD1197779

Title:

Network Sciences: Coded Computing: A Transformative Framework for Tactical Wireless Edge Computing

Author(s):

• Avestimehr, Amir S.

Author Organization(s):

• UNIVERSITY OF SOUTHERN CALIFORNIA LOS ANGELES

Report Date:

2022-03-11

Abstract:

Major Goals: Our goal in this project is to develop a foundation for the design of coded computing in order to tackle the key bottlenecks that arise in tactical wireless edge computing, namely resource (in particular bandwidth) constraints and high dynamics due to mobility and disruptions in service. In particular, our proposed research plan consists of the following main thrusts. Thrust 1: Minimum Bandwidth Coded Computing (MBCC). In this thrust we focus on the bandwidth bottleneck of large-scale tactical edge computing, and develop coded computing strategies that effectively inject computation redundancy to minimize the communication load. Thrust 2: Maximum Robustness Coded Computing (MRCC). In this thrust we consider the resiliency of tactical edge computing to network dynamics, and generalize our framework to ac- count for failing, straggling, and overloaded nodes in the network. Thrust 3: Practical Demonstrations. In parallel to the first two research thrusts, we will also design experiments that can primarily be used to test out the theory and develop optimal and near optimal resource allocation algorithms for efficient distributed edge computing.

Document Type:

Conference:

Journal:

Pages:

11

File Size:

0.16MB

Contracts:

Grants:

Descriptors:

• signal processing
• information theory
• data sets
• edge computing
• students
• machine learning
• computations
• distributed computing
• distance learning
• governments
• information processing
• network science
• scheduling (production)
• training
• abstracts
• algorithms
• bandwidth

Identifiers:

• mbcc (Minimum Bandwidth Coded Computing)
• Tactical Wireless Edge Computing
• mrcc (Maximum Robustness Coded Computing)
• Privacy-Preserving Collaborative Machine Learning
• tacc (Topology-Aware Coded Computing)
• Distributed Graph Processing
• Turbo-Aggregates
• lcc (Lagrange Coded Computing)
• Timely-Throughput Optimal Coded Computing
• Speed-Adaptive Coded Computing
• Distributed Matrix Multiplication

SubjectCategory:

• Mathematical and Computer Sciences
• Mathematical and Computer Sciences

Communities of Interest:

• Autonomy

Distribution Statement:

Approved For Public Release

View The Document