Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Synesthesia of Machines (SoM)-Enhanced ISAC Precoding for Vehicular Networks with Double Dynamics

Zonghui Yang, Shijian Gao, Xiang Cheng, Liuqing Yang

TL;DR

This work tackles ISAC precoding in vehicular networks under doubly dynamic conditions (time-varying channels and fast-moving targets). It introduces a synesthesia-of-machine (SoM) concept and a PSAC-based DRL framework to fuse multi-modal observations (CSI and position) for robust, real-time precoding via a hybrid action space. Complementing this, an optimization-based baseline with full state information and a frame-structured ISAC processing pipeline illustrate performance and complexity tradeoffs. Simulation results show DRL‑aided precoding yields notable SE and sensing improvements, with close-to-ideal performance under realistic dynamics and reduced computational burden. The approach offers a practical pathway toward timely, high‑fidelity ISAC in dynamic vehicular environments.

Abstract

Integrated sensing and communication (ISAC) technology is vital for vehicular networks, yet the time-varying communication channels and rapid movement of targets present significant challenges for real-time precoding design. Traditional optimization-based methods are computationally complex and depend on perfect prior information, which is often unavailable in double-dynamic scenarios. In this paper, we propose a synesthesia of machine (SoM)-enhanced precoding paradigm that leverages modalities such as positioning and channel information to adapt to these dynamics. Utilizing a deep reinforcement learning (DRL) framework, our approach pushes ISAC performance boundaries. We also introduce a parameter-shared actor-critic architecture to accelerate training in complex state and action spaces. Extensive experiments validate the superiority of our method over existing approaches.

Synesthesia of Machines (SoM)-Enhanced ISAC Precoding for Vehicular Networks with Double Dynamics

TL;DR

This work tackles ISAC precoding in vehicular networks under doubly dynamic conditions (time-varying channels and fast-moving targets). It introduces a synesthesia-of-machine (SoM) concept and a PSAC-based DRL framework to fuse multi-modal observations (CSI and position) for robust, real-time precoding via a hybrid action space. Complementing this, an optimization-based baseline with full state information and a frame-structured ISAC processing pipeline illustrate performance and complexity tradeoffs. Simulation results show DRL‑aided precoding yields notable SE and sensing improvements, with close-to-ideal performance under realistic dynamics and reduced computational burden. The approach offers a practical pathway toward timely, high‑fidelity ISAC in dynamic vehicular environments.

Abstract

Integrated sensing and communication (ISAC) technology is vital for vehicular networks, yet the time-varying communication channels and rapid movement of targets present significant challenges for real-time precoding design. Traditional optimization-based methods are computationally complex and depend on perfect prior information, which is often unavailable in double-dynamic scenarios. In this paper, we propose a synesthesia of machine (SoM)-enhanced precoding paradigm that leverages modalities such as positioning and channel information to adapt to these dynamics. Utilizing a deep reinforcement learning (DRL) framework, our approach pushes ISAC performance boundaries. We also introduce a parameter-shared actor-critic architecture to accelerate training in complex state and action spaces. Extensive experiments validate the superiority of our method over existing approaches.
Paper Structure (47 sections, 2 theorems, 60 equations, 15 figures, 7 tables, 2 algorithms)

This paper contains 47 sections, 2 theorems, 60 equations, 15 figures, 7 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. System Model
  3. Transmitting Signal
  4. Communication Model
  5. Sensing Model
  6. Optimization-based ISAC precoding with instantaneous state
  7. Related Works
  8. Alternating-based Optimization
  9. Optimizing $\bm F_{\text{RF}}$
  10. Optimizing $\bm F_{\text{BB},m}$
  11. Iteration
  12. Remark on optimization-based schemes
  13. Frame structure for ISAC processing in double dynamics
  14. Stage I: Precoding update
  15. Stage II: Data transmission
  16. ...and 32 more sections

Key Result

Proposition 1

There exists a global optimum for the relaxed problem satisfying the rank-1 constraint, which ensures the tightness of the SDR can be guaranteed.

Figures (15)

  • Figure 1: Investigated system model where an ISAC BS serves multiple vehicles and senses a moving target.
  • Figure 2: The frame structure of conventional, predictive, and proposed one for ISAC.
  • Figure 3: An illustration of two types of precoding update schemes ($N_{\text{t}}=8, U=4, M=2$).
  • Figure 4: An illustration of the environmental state transition.
  • Figure 5: The PSAC network architecture.
  • ...and 10 more figures

Theorems & Definitions (5)

  • Proposition 1
  • proof
  • Remark 1
  • Proposition 2
  • proof