This paper explores the integration of neuromorphic computing with wireless sensing, focusing on radar processing within the framework of Integrated Sensing and Communication (ISAC). In this context, we propose a neuromorphic signal processing module that employs an extension of the Spiking Locally Competitive Algorithm (S-LCA) to perform delay-Doppler estimation in an OFDM-based sensing setup. As sensing reference signals, we utilize sequences from a finite-dimensional Gabor frame constructed from time-frequency translates of a seed vector. We refer to a Gabor frame construction based on the Alltop seed vector due to its low mutual coherence and hardware-friendly implementation. The proposed system is deployed on the SpiNNaker neuromorphic platform, demonstrating notable power savings compared to traditional hardware.

Integrated sensing and communication (ISAC) is a cornerstone of sixth-generation (6G) wireless networks, enabling the seamless integration of high-speed communication with precise sensing and localization. The design of ISAC systems typically involves trade-offs between communication and sensing performance. This paper explores different aspects of orthogonal frequency-division multiplexing (OFDM) waveforms for monostatic radar, aiming to improve sensing performance while maintaining communication capabilities. Our derivation shows that range resolution is influenced by the shape of the baseband transmission pulse. However, as more bandwidth is allocated for sensing, the pulse's impact on resolution becomes negligible. Additionally, we investigate how resource allocation strategies affect resolution and ambiguity in both range and Doppler. Several adjacent and non-adjacent schemes are evaluated through simulations using a realistic ISAC framework developed at Fraunhofer HHI. The results highlight key trade-offs and provide recommendations for ISAC waveform design, laying a solid foundation for future research in this area.

This paper considers a general framework for massive random access based on sparse superposition coding. We provide guidelines for the code design and propose the use of constant-weight codes in combination with a dictionary design based on Gabor frames. The decoder applies an extension of approximate message passing (AMP) by iteratively exchanging soft information between an AMP module that accounts for the dictionary structure, and a second inference module that utilizes the structure of the involved constant-weight code. We apply the encoding structure to (i) the unsourced random access setting, where all users employ a common dictionary, and (ii) to the “sourced” random access setting with user-specific dictionaries. When applied to a fading scenario, the communication scheme essentially operates non-coherently, as channel state information is required neither at the transmitter nor at the receiver. We observe that in regimes of practical interest, the proposed scheme compares favorably with state-of-the art schemes, in terms of the (per-user) energy-per-bit requirement, as well as the number of active users that can be simultaneously accommodated in the system. Importantly, this is achieved with a considerably smaller size of the transmitted codewords, potentially yielding lower latency and bandwidth occupancy, as well as lower implementation complexity.

Channel charting (CC) is an emerging machine learning method for learning a lower-dimensional representation of channel state information (CSI) in multi-antenna systems while simultaneously preserving spatial relations between CSI samples. The driving objective of CC is to learn these representations or channel charts in a fully unsupervised manner, i.e., without the need for having access to explicit geographical information. Based on recent findings in deep manifold learning, this paper addresses the problem of CC via the "not-too-deep" (N2D) approach for deep manifold learning. According to the proposed approach, an embedding of the global channel chart is first learned using a deep neural network (DNN)-based autoencoder (AE), and this embedding is subsequently searched for the underlying manifold using shallow clustering methods. In this way we are able to counter the problem of collapsing extremities - a well known deficiency of channel charting methods, which in previous research efforts could only be mitigated by introducing side-information in form of distance constraints. To further exploit the ever-increasing spatio-temporal CSI resolution in modern multi-antenna systems, we propose to augment the employed AE with convolutional neural network (CNN) input layers. The resulting convolutional autoencoder (CAE) architecture is able to automatically extract sparsely distributed spatio-temporal features from beamspace domain CSI, yielding a reduced computational complexity of the resulting model.

Channel charting (CC) is an emerging machine learning framework that aims at learning lower-dimensional representations of the radio geometry from collected channel state information (CSI) in an area of interest, such that spatial relations of the representations in the different domains are preserved. Extracting features capable of correctly representing spatial properties between positions is crucial for learning reliable channel charts. Most approaches to CC in the literature rely on range distance estimates, which have the drawback that they only provide accurate distance information for colinear positions. Distances between positions with large azimuth separation are constantly underestimated using these approaches, and thus incorrectly mapped to close neighborhoods. In this paper, we introduce a correlation matrix distance (CMD) based dissimilarity measure for CC that allows us to group CSI measurements according to their co-linearity. This provides us with the capability to discard points for which large distance errors are made, and to build a neighborhood graph between approximately collinear positions. The neighborhood graph allows us to state the problem of CC as an instance of an Euclidean distance matrix completion (EDMC) problem where side-information can be naturally introduced via convex box-constraints.