Wireless Multiaccess Distributed Computing Networks

doi:10.1109/JIOT.2026.3674652

Wireless Multiaccess Distributed Computing Networks

2026-03-16

0

PRE

AI

DOI：10.1109/JIOT.2026.3674652

Original

Original request for help

Save

Abstract

En 中文

Wireless distributed computing (WDC) is a promising architecture for enabling high computing efficiency in future Internet of Things (IoT) systems. In WDC networks, the Shuffle phase suffers from high normalized delivery time (NDT) due to two main factors, i.e., heavy communication load and low communication efficiency. Current research mainly focuses on improving communication efficiency, often overlooking the need to reduce the communication load. In this article, from the perspective of jointly reducing the communication load and improving communication efficiency, we propose a wireless multiaccess distributed computing (WMADC) network and a novel multiaccess interference alignment and neutralization (MA-IAN) scheme to reduce the NDT. In particular, the WMADC network is proposed to greatly reduce the communication load through transmitting fewer intermediate values (IVAs) due to the advantage that a reducer node can be connected with multiple mapper nodes and directly obtain their computed IVAs. Furthermore, we propose the novel MA-IAN scheme to greatly improve communication efficiency by carefully designing the sophisticated joint interference alignment (IA) and interference neutralization (IN) scheme. Finally, we derive an information-theoretic lower bound on the NDT for the WMADC network, which demonstrates that the proposed MA-IAN scheme is optimal in several scenarios. Furthermore, the proposed WMADC network with the MA-IAN scheme outperforms existing relevant schemes in terms of reducing the NDT.

Keywords:

Interference alignment and neutralization (IAN)

multiaccess distributed computing (MADC)

normalized delivery time (NDT)

wireless communication

Wireless Multiaccess Distributed Computing Networks

Abstract

Journal

Organization