Subject: Cutting-Edge Advancements in Communication and Sensing

Hi Elman,

This newsletter covers recent breakthroughs in communication and sensing, highlighting the growing influence of deep learning, innovative hardware developments, and novel signal processing techniques.

General Overview

This collection of papers explores advancements across various domains within communication and sensing, from cost-effective hardware solutions to sophisticated signal processing techniques and novel applications. A notable theme is the increasing integration of deep learning for tasks ranging from channel prediction and interference mitigation to semantic communication and health monitoring. For instance, De Filippo et al. (2025) De Filippo et al. (2025) introduce a hybrid CNN-LSTM architecture for channel prediction in 6G non-terrestrial networks, aiming to reduce pilot overhead. Concurrently, multiple papers address the challenges of massive MIMO precoding, with Kasalaee et al. (2025) Kasalaee et al. (2025) focusing on compression of site-specific deep neural networks for improved energy efficiency and Karkan et al. (2025) Karkan et al. (2025) proposing a low-complexity plug-and-play deep learning model with enhanced generalization across different sites. These works highlight the ongoing effort to make computationally demanding deep learning approaches practical for real-world deployment in complex communication systems.

Another significant trend is the development of innovative hardware and system architectures for enhanced sensing and communication capabilities. Euchner et al. (2025) Euchner et al. (2025) present ESPAGOS, an ultra low-cost WiFi channel sounder, democratizing access to channel measurements for various applications, including Joint Communications and Sensing (JCaS). Tiwari and Caire (2025) Tiwari & Caire (2025) explore array-fed Reflective Intelligent Surfaces (RIS) for flat-top beamforming, offering a more energy-efficient alternative to traditional active array designs. Furthermore, Xu et al. (2025) Xu et al. (2025) introduce a novel pinching antenna system (PASS) and propose both optimization-based and learning-based methods for joint transmit and pinching beamforming, demonstrating significant performance gains over conventional massive MIMO. These contributions showcase the exploration of novel hardware paradigms to address the evolving demands of future communication systems.

Several papers delve into specific signal processing challenges and applications within the broader communication and sensing landscape. Zheng et al. (2025) Zheng et al. (2025) employ an attention mechanism-guided GAN for removing motion artifacts in opto-physiological monitoring, while Werner et al. (2025) Werner et al. (2025) develop a cost-effective Direction of Arrival (DoA) estimation system using SDRs and a switched uniform circular array. In the realm of molecular communication, Cai and Akan (2025) Cai & Akan (2025) propose a semantic learning framework for enhanced diagnostic tasks in the Internet of Bio-Nano Things (IoBNT). These diverse applications demonstrate the breadth of research within the field and the potential impact of these technologies across various domains.

Beyond specific applications, foundational work on modulation and multiple access techniques is also presented. Deng et al. (2025) Deng et al. (2025) provide a unifying view of Orthogonal Time Frequency Space (OTFS) modulation and its variants, while Yan et al. (2025) Yan et al. (2025) introduce ReMAC, a novel digital multiple access computing scheme based on repeated transmission. Aditya et al. (2025) Aditya et al. (2025) discuss the potential of Rate-Splitting Multiple Access (RSMA) for 6G, presenting both simulation and experimental results. These contributions highlight the ongoing development of advanced modulation and multiple access schemes to meet the increasing demands for spectral efficiency and reliable communication in complex environments.

Finally, several papers address practical challenges related to system implementation and performance evaluation. Kuzio et al. (2025) Kuzio et al. (2025) propose a procedure for assessing the quality of machine health index data prediction, crucial for predictive maintenance applications. Tapparel and Burg (2025) Tapparel & Burg (2025) investigate LoRa fine synchronization with a two-pass time and frequency offset estimation method. These works emphasize the importance of robust and reliable methods for evaluating and improving the performance of real-world systems. Collectively, these papers offer a snapshot of the current research landscape, highlighting the ongoing efforts to develop innovative solutions for next-generation communication and sensing systems.

Paper Highlights

MovISAC: A Breakthrough in Moving Target Imaging

MovISAC: Coherent Imaging of Moving Targets with Distributed Asynchronous ISAC Devices by Jacopo Pegoraro, Dario Tagliaferri, Joerg Widmer https://arxiv.org/abs/2502.08236

Caption: This figure illustrates the MovISAC processing chain, highlighting the key steps involved in achieving high-resolution imaging of moving targets. The process begins with over-the-air synchronization and single-pair image formation, followed by Doppler estimation and a novel Doppler-space association algorithm to determine target velocities. Finally, Doppler pre-compensation and coherent slow-time image formation produce focused images of the moving targets.

Integrated sensing and communication (ISAC) networks hold immense potential for high-resolution imaging, exceeding the capabilities of individual devices. However, imaging moving targets with existing coherent imaging techniques is challenging due to the Doppler effect, which introduces blurring and inaccurate positioning. Traditional methods for moving target imaging, commonly used in synthetic aperture radar (SAR) and inverse SAR (ISAR), rely on computationally intensive exhaustive velocity searches, rendering them impractical for real-time applications. MovISAC tackles these limitations by introducing a novel approach to coherent imaging of moving targets using distributed asynchronous ISAC devices.

At the heart of MovISAC is its innovative Doppler-space association algorithm. After performing over-the-air synchronization to correct timing, frequency, and phase offsets among the distributed ISAC devices, MovISAC generates low-resolution images from each device pair. It then estimates the Doppler shifts from these images and coarsely localizes the targets. The association algorithm ingeniously matches the observed Doppler shifts with the corresponding spatial peaks, enabling precise estimation of each target's velocity vector: v_q = U^†(x_q)A(q) , where v_q represents the velocity vector of target q, U(x_q) is the velocity regression matrix, and A(q) is the association function. This unique capability allows MovISAC to pre-compensate for the Doppler shift of each target individually before forming the final high-resolution images.

This pre-compensation effectively transforms the Doppler effect from a hindrance into an asset. By selectively compensating only for the Doppler shifts of the desired targets, MovISAC coherently cancels the signatures of other moving targets and static clutter, producing target-specific high-resolution images. This clutter resilience significantly enhances target resolution and localization accuracy. The final image after slow-time summation is given by: Î_q(x) ≈ ξ_qKe^−jθqH(x − x_q) + Z(x), where Î_q(x) is the final image of target q, ξ_q and θ_q are the amplitude and scattering phase of the target, H(x) is the spatial ambiguity function of the ISAC network, and Z(x) is the noise.

Numerical simulations confirm the superior performance of MovISAC. Compared to existing methods like standard multistatic imaging (SMI) and iterative SAF subtraction (ISAFS), MovISAC achieves up to 18 times lower localization error, with median RMSE values around 0.5 cm. Moreover, it enables accurate velocity estimation with cm/s-level accuracy, a feature absent in current imaging methods. The simulations also highlight MovISAC's ability to resolve targets separated by as little as 1 cm, effectively reaching the theoretical resolution limit. Importantly, MovISAC achieves this performance with significantly lower complexity compared to exhaustive velocity search methods, making it suitable for real-time applications.

RSMA: Redefining Multiple Access for 6G

Rate-Splitting Multiple Access for 6G: Prototypes, Experimental Results and Link/System level Simulations by Sundar Aditya, Yong Jin Daniel Kim, David Vargas, David Redgate, Onur Dizdar, Neil Bhushan, Xinze Lyu, Sibo Zhang, Stephen Wang, Bruno Clerckx https://arxiv.org/abs/2502.09283

Caption: Link-level simulations comparing the Block Error Rate (BLER) performance of Rate-Splitting Multiple Access (RSMA) and Space Division Multiple Access (SDMA) for two users in an enhanced Mobile Broadband (eMBB) use case. RSMA demonstrates approximately a 3dB SNR gain over SDMA across various modulation and coding schemes (represented by different colored lines and markers).

Rate-Splitting Multiple Access (RSMA) is emerging as a powerful physical layer multiple access technique for 6G, potentially surpassing the capabilities of 5G's Space Division Multiple Access (SDMA). RSMA offers superior interference management by partially decoding interference and partially treating it as noise, adapting to channel conditions more effectively than SDMA or Non-Orthogonal Multiple Access (NOMA). This adaptability translates to improved spectral efficiency and fairness, making RSMA a promising successor to SDMA in the evolution towards 6G. The article presents simulation and experimental results demonstrating RSMA's advantages in realistic deployment scenarios for key 6G use cases like enhanced Mobile Broadband (eMBB) and Integrated Sensing and Communications (ISAC).

Link-level simulations, adapting the 5G New Radio (NR) Physical Downlink Shared Channel (PDSCH) to implement RSMA, showed a 3dB SNR gain in block error rate (BLER) over SDMA for the eMBB use case. These simulations also revealed that RSMA's performance remained robust across different receiver architectures, including low-complexity non-SIC receivers, making them viable alternatives to SIC-based receivers. Further analysis showed that RSMA's gains are most pronounced when co-scheduled users are in close proximity (less than 20m apart).

System-level simulations in realistic indoor hotspot (InH) and urban microcell (UMi) environments further validated RSMA's superiority. The results demonstrated that RSMA achieves greater fairness at higher sum rates compared to SDMA, particularly for users with small SINR disparities and high spatial correlations (typically, closely located users). For instance, in the UMi scenario with four users, RSMA offered user-rate gains exceeding 100% with zero-forcing precoders, highlighting its robustness to high-interference scenarios.

Experimental results from two independently built RSMA prototypes further solidified the simulation findings. In downlink unicast communication to two users, RSMA demonstrated higher sum throughput with better fairness than both SDMA and NOMA across various channel conditions. In overloaded MIMO scenarios (number of users exceeding the number of transmit antennas), RSMA outperformed SDMA with user scheduling by serving all users simultaneously, leading to a consistent minimum throughput over time. For ISAC scenarios involving simultaneous communication and sensing, RSMA without a dedicated sensing signal achieved the largest performance envelope in terms of both throughput and radar SNR, particularly in high-interference scenarios with significant overlap between sensing and communication directions. These findings, coupled with ongoing standardization efforts in 3GPP and ETSI, position RSMA as a strong contender for the dominant multiple access technology in 6G.

Semantic Communication in Heterogeneous Networks: Addressing the Challenges

Semantic Communication Meets Heterogeneous Network: Emerging Trends, Opportunities, and Challenges by Guhan Zheng, Qiang Ni, Aryan Kaushik, Lixia Yang https://arxiv.org/abs/2502.08999

Caption: Comparison of MSE between the proposed framework and traditional aggregation.

Next-generation wireless networks promise groundbreaking advancements, but challenges persist in achieving high spectral efficiency and robustness. Semantic Communication (SemCom) emerges as a potential solution by shifting the focus from transmitting raw data to transmitting meaning. By employing ML-based semantic codecs, SemCom reduces transmission volume and increases robustness against channel impairments. However, the task-oriented nature of these codecs introduces a unique challenge: they require frequent updates, especially in heterogeneous networks with diverse users and dynamic conditions. Unlike simple point-to-point communication, network-wide updates involve multiple users relying on a shared, evolving semantic understanding, significantly increasing the complexity of maintaining consistency.

The key components of a SemCom system include the semantic knowledge base, which stores background knowledge, and the semantic encoder/decoder pair. The knowledge base needs synchronization across users to adapt to new tasks, while the encoder extracts meaning and the decoder reconstructs it. The semantic channel, analogous to the physical channel, focuses on semantic integrity and fidelity. However, several key challenges arise in heterogeneous networks. System heterogeneity stems from the diverse computational and communication capabilities of users, while codec heterogeneity arises from the varied initial codec models employed by different users. The personalized one-to-many model challenge arises from the need for a single encoder to serve multiple heterogeneous decoders, and data heterogeneity further complicates matters due to the non-IID nature of user datasets.

The paper proposes a heterogeneous semantic codec updating framework to tackle these challenges. This framework involves users assessing each other's trustworthiness, electing a central node for aggregation, providing random data for a global dataset, performing local training to evaluate convergence and accuracy, and finally uploading updated models for weighted aggregation by the central node. This framework accounts for the heterogeneity of user models and data distributions, using adaptive re-weighting strategies to improve model generalization.

Testing this framework with existing SemCom codec models demonstrated its effectiveness. Using two different models and a CIFAR-10 dataset, the results showed that traditional aggregation methods led to lower convergence and accuracy for one of the models. However, the proposed framework, by incorporating mutual understanding and reassigning aggregation weights, improved both convergence speed and training accuracy for both models.

Despite the potential of SemCom, open research questions remain. These include addressing potential discrimination against users with slower updates, ensuring fairness in data representation, protecting the privacy of personalized models, securing the system against evolving attack strategies, adapting to dynamic network conditions with joining and leaving users, and motivating user participation in updates. These challenges highlight the need for further research to fully realize the potential of SemCom in the complex landscape of heterogeneous networks.

Conclusion

This newsletter showcases exciting advancements in communication and sensing. MovISAC's innovative approach to imaging moving targets using distributed ISAC devices offers significant improvements in resolution and accuracy, potentially revolutionizing applications like autonomous driving and surveillance. RSMA presents a compelling alternative to SDMA for 6G, demonstrating superior performance in various scenarios, including eMBB and ISAC, paving the way for more efficient and fair multiple access. Finally, the exploration of semantic communication in heterogeneous networks highlights the transformative potential of transmitting meaning rather than raw data, while also revealing the complex challenges in updating and maintaining shared semantic understanding across diverse users. These advancements collectively represent a significant step towards realizing the vision of next-generation communication and sensing systems.