Quantum Integrated Communication and Computing Over Multiple-Access Bosonic Channel
2026-04-09 • Information Theory
Information Theory
AI summaryⓘ
The authors study a way to use quantum signals to both compute and communicate information at the same time on a special channel where multiple users send data together. They design a system where one receiver can do calculations on analog signals from some devices while also decoding digital messages from others, using the quantum properties of the channel. To make this work efficiently, they create a method to adjust sending power and receiving settings that balances computation accuracy with communication speed, solving a complex math problem with a new approach that is both fast and simple.
quantum multiple-access channelbosonic channelcoherent-state signallingover-the-air computationpower controlreceive coefficientsum-rate constraintnon-convex optimizationminimum-mean square errorprojected-gradient method
Authors
Ioannis Krikidis
Abstract
We investigate a quantum integrated communication and computation (QICC) scheme for a single-mode bosonic multiple-access channel (MAC) with coherent-state signalling. By exploiting the natural superposition property of the quantum MAC, a common receiver simultaneously performs over-the-air computation (OAC) on the analogue symbols transmitted by one set of devices and decodes multiple-access data from another. The joint design of the transmit power control and the receive coefficient leads to a non-convex optimization problem that maximizes computation accuracy under a prescribed sum-rate communication constraint. To address this challenge, we develop a low-complexity alternating-optimization framework that incorporates: (i) closed-form linear minimum-mean square error updates for the receive coefficient, (ii) monotonicity properties of the quantum sum-rate constraint, and (iii) projected-gradient refinements for the communication powers. The proposed QICC scheme achieves an effective computation-communication trade-off with fast convergence and low computational complexity.