Project Description

Sub-Project 1: ISAC Scheduling and Networking for Perceptive Wireless Networks

To realize large-scale perceptive wireless networks, we must go beyond evaluating small-scale ISAC systems and address performance at the network level, raising key questions:

• How does system performance scale with network size?

• How can we design optimal resource allocation and task scheduling to achieve this?

• How do we accommodate heterogeneous users and devices with diverse sensing and communication needs?

• How can AI enhance the intelligence and scalability of such networks?

This subproject aims to tackle these challenges. Traditionally, sensing has supported communication, but in peceptive wireless networks, Sensing-as-a-Service (SaaS) becomes central. In many smart applications, sensing Quality-of-Service (QoS) may be as important, or more so, than communication QoS. As such, optimizing ISAC to meet these requirements is a key issue.

While existing research has addressed ISAC physical layer techniques (e.g., waveform and beamforming), fundamental trade-offs between sensing and communication—like increased interference when boosting sensing range—remain critical, especially in large-scale or random networks. Most studies focus on small-scale settings, leaving scaling laws, interference control, and optimal transmission strategies in large networks underexplored.

Moreover, Device-to-Device (D2D) communication offers advantages such as lower latency and higher spectral efficiency, but incorporating sensing needs adds complexity. While prior work addressed beamforming and interference, joint scheduling for communication and sensing in D2D remains unresolved.

Project objectives include:

1. Analyze the fundamental tradeoff for large-scale ISAC networks in homogeneous settings, derive scaling laws, and develop optimal transmission and sensing designs.

2. Extend to heterogeneous networks with diverse communication/sensing demands and develop fair and efficient scheduling for both D2D and infrastructure-based networks.

3. Incorporate AI to manage complex resource allocation and scheduling, enabling intelligent and scalable network designs.

Sub-Project 2: ISAC Cross-Layer Transceiver Design and Resource Management for AI-Empowered Perceptive Wireless Networks

To realize various functions of perceptive networks, the design of the physical layer and medium access control layer is critical. Although various transmission and reception technologies and system designs for ISAC are currently under active discussions, the existing discussions are still incomplete. Moreover, as applications related to AI and the Internet of Things (IoT) increasingly demand more uplink transmission than downlink, the design of uplink-downlink balanced ISAC systems and networks has become essential. In this context, the goals of this subproject primarily focus on the research and development of AI-enabled cross-layer system designs for ISAC in perceptive wireless networks. It aims to provide relevant physical-layer and cross-layer designs that can meet the quality of service requirements for various communication and sensing tasks. The key technologies of this sub-project include:

• Downlink and uplink ISAC transceiver designs

• Energy-efficient ISAC transceiver system design

• Multi-base station and multi-point ISAC system design

• Dynamic uplink and downlink ISAC resource allocation techniques

• QoS-based adaptive ISAC system design methods

• AI-based low-complexity, high-efficiency ISAC system design methods

Sub-Project 3: Context Reasoning and Optimization Strategy for Coordinated ISAC

Sub-project 3 primarily focuses on Coordinated ISAC, also known as networked or cooperative ISAC, to enhance coordination gain, which not only comes from improved link-level performance, but also emphasizes on strengthening system-level and even vertical application performance. Therefore, Coordinated ISAC needs to be able to fuse heterogeneous information from underlying sensing and communication layers and coordinate multiple network nodes. This coordination is crucial for formulating strategies that optimize system-level or vertical application performance.

To achieve this goal, Coordinated ISAC must be able to accomplish the following:

1. Context Inference: This involves inferring more than just channel models. It extends to inferring the network deployment environment (e.g., indoor or outdoor, coverage area), the geographical structure of the deployment environment, signal propagation patterns within that environment, mobile terminal movement patterns, potential information security and privacy risks, vertical application service requirements and goals, base station movement patterns, possible changes in the deployment environment, and interactions between multiple network nodes.

2. Performance Optimization Strategy Formulation and Learning: After performing context inference, the next step is to formulate performance optimization strategies. These strategies will dictate how each network layer operates and is configured for every network node. Because sensing and communication involve diverse heterogeneous information, the performance optimization problem might not even be expressible in an analytical form, making it intractable for traditional optimization methods. Therefore, learning performance optimization strategies through machine learning algorithms becomes a more feasible approach.

Consequently, Sub-project 3 will focus on context inference and performance optimization strategy learning in Coordinated ISAC to optimize performance at the system level or vertical application level.

Sub-Project 4: ISAC Signal Processing Techniques and Array Architecture Design for Perceptive Wireless Networks

The parallel and mostly independent development of sensing and communication in the past few decades has led to duplications of hardware and signal processing methods. In particular, there are similarities between sensing and communication in terms of array signal processing. For instance, the concept of phased-MIMO radar sensing is similar to that of hybrid analog and digital MIMO communications. The combined use of millimeter waves (mmWaves) and massive MIMO also makes mmWave MIMO communication channel estimation resemble two-dimensional sensing angle estimation. These similarities result in a great opportunity for integrating sensing and communication. 

In this subproject, ISAC signal processing techniques and array architecture design for perceptive wireless networks are studied. Joint sensing and communication performance optimization is also tackled. The goal is to enhance sensing and communication performance while reducing hardware and computational complexity. Several advanced array signal processing techniques, including hybrid analog and digital signal processing (for reducing hardware cost and power consumption), beamspace processing (for lowering computational complexity of angle/channel estimation), as well as sparse arrays and super-resolution angle estimation methods (for enhancing estimation performance), are also explored for potentially better ISAC design. 

The specific objectives of this subproject include: 

1. Develop multi-user single-input-multiple-output (MU-SIMO) ISAC signal processing techniques and array architecture design in homogeneous networks.

2. Develop MIMO ISAC signal processing techniques and array architecture design in heterogeneous networks.

3. Develop AI-assisted ISAC signal processing techniques and array architecture design to further improve performance.

Sub-Project 5: Secure AI-RAN Empowered ISAC Management Platforms

To support scalable and intelligent wireless networks in 6G environments, it is necessary to integrate communication and sensing with native artificial intelligence, while ensuring secure and efficient system management. This subproject addresses key challenges related to system-level orchestration in AI-RAN enabled ISAC (Integrated Sensing and Communication) networks, prompting the following questions:

• How can real-time sensing and communication be jointly managed across heterogeneous network environments?

• What strategies enable efficient deployment and decision-making under constraints of latency, spectrum, and privacy?

• In what ways can system architectures incorporate satellite, terrestrial, and indoor networks into a unified management plane?

• How can federated learning and data-driven inference contribute to secure, adaptive, and interoperable network behavior?

This subproject seeks to advance system management for next-generation networks, where AI-based control is embedded in the core architecture. The focus shifts from signal-level optimization to cross-domain integration of sensing, communication, and decision-making processes. While prior work has made progress in algorithmic intelligence and RAN modularization, comprehensive frameworks that address orchestration, deployment scalability, and layered security remain underdeveloped.

In particular, the project targets indoor GPS-free and outdoor GPS-constrained environments in the FR1 (sub-6GHz) spectrum, with future extensions to FR2 and FR3 bands. These environments demand tight integration of real-time beamforming, network-level context recognition, and security enforcement across devices and domains. Moreover, the inclusion of LEO satellite communication further complicates synchronization, trust management, and edge AI operations.

The proposed platform will be evaluated as a vertically integrated testbed, incorporating key results from other subprojects at the signal, network, and application levels. It aims to validate compliance with 3GPP and O-RAN specifications, while addressing performance bottlenecks in dynamic scheduling, low-latency inference, and cross-layer system resilience.

Project objectives include:

1. Design a unified system architecture for scalable ISAC platforms that integrates sensing, communication, and AI-based control under security constraints.

2. Develop deployment strategies across FR1–FR3 spectrum ranges, incorporating satellite-terrestrial handoff, indoor-outdoor interoperability, and application-level service management.

3. Construct and evaluate a demonstration platform to validate real-time orchestration, data fusion, and compliance with evolving AI-RAN and ISAC standards.