In the rapidly evolving landscape of technology, sensor networks and the Internet of Things (IoT) have emerged as transformative forces, revolutionizing the way we interact with and understand our physical world. As we transition from the era of 4G to the dawn of 6G wireless networks, a fundamental rethinking of the purpose of radio signals has given rise to the concept of Integrated Sensing and Communications (ISAC). This pivotal domain of research and development promises to usher in a new era of connectivity, where communication extends beyond mere data transfer to encompass sensing, knowledge, intelligence, and reconfiguration, seamlessly bridging the physical and digital realms.
Limitations of Existing ISAC Approaches
Despite the significant progress made in the fundamentals of ISAC, the current vision often falls short in addressing several critical components. Firstly, it lacks support for widely distributed deployments necessary to track both connected user equipment (UEs) and passive objects over extended time and space. Secondly, it focuses solely on performance indicators, neglecting the comprehensive perspective that 6G must adopt, which includes both Key Performance Indicators (KPIs) and Key Value Indicators (KVIs). Lastly, the existing ISAC models focus only on the 6G signal as a sensor, ignoring the fusion opportunities with external sensors and more general semantic awareness.
Introducing the DISAC Framework
To unlock the full potential of ISAC, we propose an extended vision termed Distributed and Intelligent Integrated Sensing and Communications (DISAC). The DISAC framework represents a transformative approach for 6G wireless networks, built upon three interrelated cornerstones:
1. DISAC Architecture
The DISAC architecture serves as the foundation for both sensing and communications, while simultaneously offering support for intelligent operations and distributed functions. This distributed aspect not only enables large-scale tracking of connected UEs and passive objects but also revolutionizes the fundamental fabric of wireless networks. The DISAC architecture supports distributed Artificial Intelligence (AI) operations, balancing local data processing and fusion, leveraging novel multi-antenna technologies, and incorporating an exposure framework for external sensors.
2. Semantic and Goal-Oriented Framework
The semantic and goal-oriented framework, supported by Machine Learning (ML) and AI, provides an intelligent and parsimonious approach encompassing sensing activation, waveform design, signal processing, dedicated resource allocation, robust protocols, and semantic reasoning about multi-modal sensed information. This framework ensures exceptional sensing performance for a myriad of use cases while optimizing resource utilization.
3. Advanced High-Resolution Processing
The advanced high-resolution processing leverages the massively distributed observations, balancing computational and storage requirements. By exploiting multi-site and multi-band processing, and by combining ML and model-based signal processing, efficient methods that can support the goal-oriented semantic framework while running over the DISAC architecture must be developed and verified.
Enabling Technologies and Standardization Efforts
Deploying DISAC on a global scale requires standardization efforts, especially concerning AI/ML-driven sensing within distributed heterogeneous architectures. This would impact various layers of the 6G ecosystem, including the physical layer, the control and management planes, and the security of data and model exchanges. Achieving harmonious orchestration of radio, transport, and processing resources becomes of primary importance, demanding efficient and dynamic solutions.
Several standards development organizations (SDOs) have begun to explore and develop work on the ISAC topic in recent years, integrating AI-based approaches with varying degrees of progress:
- ETSI: Established a new Industry Specification Group (ISG) dedicated to ISAC, with a focus on channel modeling and complementary subjects beyond the current 3GPP scope.
- 3GPP: Completed a feasibility study on ISAC use cases and requirements, and started the specification work covering sensing operation, functional requirements, security, and secrecy aspects.
- ITU-R WP5D: Considers ISAC as a new usage scenario to enable innovative services and solutions, where the semantic nature of DISAC is expected to play a pivotal role.
- IEEE 802.11: Launched the 802.11bf Task Group, defining physical and MAC layer modifications to enhance sensing operations of wireless local-area networks in unlicensed frequency bands.
In addition to the SDO efforts, feasibility studies and normative work on AI/ML have already begun at 3GPP and IEEE, focusing on enhancements to data collection, signaling, and efficient training data management.
The Semantic Cornerstone of DISAC
The DISAC vision relies on the semantic and goal-oriented communications approach as a crucial component, allowing diverse information from various sensing modalities to be aggregated and facilitating the transition from traditional data fusion to the composition of semantically selected information and the pragmatic generation of AI-based reasoning stimuli.
Sensor Networks leveraging semantic and goal-oriented communications for distributed ISAC can provide advantages in terms of interoperability, contextual understanding, and effectiveness in activating sensing functions. This includes context- and application-adapted waveform design, signal processing, dedicated resource allocation, robust protocols, and reasoning about multi-modal sensed information.
By establishing and exploiting connections or relationships between different pieces of information derived from sensed and already available data, based on causal and semantic representation extraction, communications, and reasoning-based interpretation and composition, DISAC can achieve a unified understanding that is instrumental to effectively analyze, interpret, and derive insights from data, ultimately optimizing the use of sensing, computation, and communication resources.
Advancing the DISAC Physical Layer
The DISAC physical layer involves four key components of the classical signal processing:
- Waveform optimization: Optimizing resource allocation across time, frequency, and spatial dimensions to maximize ISAC performance.
- Channel parameter estimation: Extracting position-related parameters, such as range, angle, and Doppler shift, from received waveforms.
- Detection, data association, and estimation: Detecting objects, associating data, and estimating device positions based on fused information from different receivers.
- Tracking: Tracking UEs and objects over time.
Extending these classical signal processing principles to the DISAC context is challenging, as it requires revisiting these components to tailor them to the instantaneous application needs, completely changing the optimization and learning approaches. If executed correctly, this will lead to a more parsimonious ISAC operation.
To reach its full potential, the inclusion of semantics as a cornerstone of DISAC requires a dedicated semantic plane, enabling semantic-aware encoding/decoding over multiple frequency bands to increase the performance and robustness of sensing and communications. Additionally, flexible coordination among transmitted waveforms, such as those based on OFDM or learned from scratch, can provide improved efficiency, resilience, and robustness against interference and hardware distortion.
On the receiver-side processing, a combination of ML (e.g., for imaging and extended target processing) and model-based methods (e.g., for deciding which sensors to activate and when) is foreseen to address the challenge of aligning the physical layer with the semantic framework.
Harnessing the Power of Distributed Aperture
Resolution is a critical factor for all sensing applications, as it determines the ability to separate and distinguish objects. In DISAC, an additional dimension of distributed aperture is available, providing high resolution on the order of the wavelength while suppressing sidelobes through the random placement of the many distributed receivers.
Realizing and harnessing this distributed aperture resolution is challenging, requiring extreme calibration of the infrastructure, both in terms of the static locations of access points (APs) and the dynamic time, frequency, and phase synchronization. Depending on the level of coherence/synchronization between different APs, different high-resolution methods should be applied, and varying amounts of data should be shared.
Methods from distributed Multiple-Input Multiple-Output (MIMO) radar and imaging in Synthetic Aperture Radar (SAR) are expected to play an important role in addressing these challenges, along with the consideration of the large diversity of MIMO technologies, such as Reconfigurable Intelligent Surfaces (RISs), Distributed MIMO (D-MIMO), and Extremely Large-Scale MIMO (XL-MIMO), as well as their hardware models and impairments.
Sensing-Aided Communications for Efficient Network Operations
The DISAC paradigm envisions the optimization of large-scale and distributed communication provisioning, with location and more generally context information becoming available from the distributed processing of data gathered from multi-modal sensory devices integrated within the wireless network infrastructure. This paves the way for more efficient, context-aware, and environmentally sustainable communication networks.
Even in simple scenarios, sensing-aided communication can offer significant performance gains. Leveraging direction estimation in the uplink can lead to around a 25% increase in the achievable spectral efficiency for the downlink, as shown in the example provided in the source information.
Adaptive Resource Allocation: The Key to Harmonizing Sensing and Communications
The development of adaptive resource allocation schemes is fundamental within the DISAC concept. These techniques need to further enable simultaneous sensing and communications in dynamically evolving scenarios with heterogeneous nodes, accounting for the unique demands of high-resolution sensing leading to potentially large volumes of sensor data and semantic processing.
Unlike traditional resource allocation strategies, which primarily focus on bandwidth and power constraints to provide communication services, DISAC-oriented resource allocation must account for the unique demands of high-resolution sensing, leading to potentially large volumes of sensor data and semantic processing, which will reduce these data volumes intelligently over time and space.
Efficient resource allocation schemes must be inherently goal-oriented, exploiting contextual information to strike a balance in the time, frequency, and energy resources allocated for sensing and communications. This dynamic adjustment to varying network conditions and user requirements is crucial, especially considering the heterogeneous nature of future wireless networks.
By incorporating advanced signal processing methods and the novel semantic domain within DISAC, the resulting adaptive resource allocation strategies will ensure that both sensing data and communication signals are processed and transmitted efficiently, addressing not only the integrity or timeliness of the data but also the relevance of the content, prioritizing critical information for respective transmission.
DISAC Architecture: Enabling Distributed Intelligent Operations
The DISAC architecture will depart from the conventional cellular network architecture in several ways, supporting distributed intelligent signal processing at the sensing receive nodes so that only higher-level sensing information is sent to the fusion center. This involves defining new network functions, protocols, and interfaces, as well as new functions dedicated to tracking and handover of objects, requiring the definition and management of SRN groups and suitable identifiers for objects without subscriber identity module (SIM) cards.
The DISAC architecture must support a variety of heterogeneous capabilities and be able to balance the load based on the instructions from the semantic layer to optimally perform local processing and hierarchical fusion of information. It needs to support distributed functions across many nodes, including synchronization, calibration, semantic extraction, semantic composition, and semantic instruction, as well as relevant AI modules at both the semantic and radio layers.
Unlocking the Potential of DISAC
By seamlessly integrating sensing with communications and harnessing the power of distributed AI, DISAC not only addresses the limitations of existing ISAC approaches but also unlocks new possibilities for resource-efficient, accurate, and semantic network operations.
However, the challenges in realizing the DISAC vision are formidable, ranging from designing and executing the semantic framework to developing new hardware components and addressing real-world demonstration of the semantic framework. Standardization efforts are also crucial, as studies on waveforms, distributed architectures, and specific metrics and KPIs applicable to DISAC must be discussed in SDOs.
As the sensor networks and IoT landscape continues to evolve, the DISAC framework represents a transformative approach that can unlock the true potential of integrated sensing and communications, paving the way for a future of smarter, more efficient, and context-aware wireless networks. By harnessing the power of distributed intelligence, the DISAC vision promises to revolutionize how we collect, process, and act upon the wealth of information available in our physical world.