Sebastian Cammerer
Research Subjects
Sebastian Cammerer has a total of 56 co-authors. Here are the top co-authors by number of shared publications:
- Stephan Ten Brink
Institute of Telecommunications, University of Stuttgart, Stuttgart, Germany
33 shared publications - Ahmed Elkelesh
Institute of Telecommunications University of Stuttgart Pfaffenwaldring 47 70569 Stuttgart Germany
10 shared publications - Moustafa Ebada
Institute of Telecommunications University of Stuttgart Pfaffenwaldring 47 70569 Stuttgart Germany
10 shared publications - Sebastian Dörner
Institute of Telecommunications University of Stuttgart Pfaffenwaldring 47 70659 Stuttgart Germany
10 shared publications - Jakob Hoydis
Nokia Bell Labs, Paris-Saclay, 91620 Nozay, France
7 shared publications Alexander Keller
University of Stuttgart
6 shared publications
Latest Publications
- 31 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications
- Vol. 32 (6) , 11-14
AI-Native 6G: Empowering Intelligent RAN With Accelerated Compute
The coming era of 6G will be marked by a paradigm shift in the radio access network (RAN), propelled by pervasive and embedded artificial intelligence (AI). In this article, we examine the essential roles of software-defined, accelerated compute platforms with graphics processing units (GPUs) in building AI-native 6G RAN.
- 31 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Wireless Communications
- Vol. 32 (6) , 11-14
AI-Native 6G: Empowering Intelligent RAN With Accelerated Compute
The coming era of 6G will be marked by a paradigm shift in the radio access network (RAN), propelled by pervasive and embedded artificial intelligence (AI). In this article, we examine the essential roles of software-defined, accelerated compute platforms with graphics processing units (GPUs) in building AI-native 6G RAN.
- Open Access
Serial Polar Automorphism Ensemble Decoders for Physical Unclonable Functions
Physical unclonable functions (PUFs) involve challenging practical applications of error-correcting codes (ECCs), requiring extremely low failure rates on the order of $10^{-6}$ and below despite raw input bit error rates as high as 22%. These requirements call for an efficient ultra-low rate code design. In this work, we propose a novel coding scheme tailored for PUFs based on Polar codes and a low-complexity version of automorphism ensemble decoding (AED). Notably, our serial AED scheme reuses a single successive cancellation (SC) decoder across multiple decoding attempts. By introducing cascaded and recursive interleavers, we efficiently scale the number of AED candidates without requiring expensive large multiplexers. An aggressive quantization strategy of only 3 bits per message further reduces the area requirements of the underlying SC decoder. The resulting coding scheme achieves the same block error rate of $10^{-6}$ as our baseline based on Bose-Ray-Chaudhuri-Hocquenghem (BCH) codes while requiring 1.75x fewer codeword bits to encode the same K = 312 payload bits. This reduction translates directly into 1.75x less helper data storage and, consequently, a smaller overall chip area.Open AccessSerial Polar Automorphism Ensemble Decoders for Physical Unclonable Functions
Physical unclonable functions (PUFs) involve challenging practical applications of error-correcting codes (ECCs), requiring extremely low failure rates on the order of $10^{-6}$ and below despite raw input bit error rates as high as 22%. These requirements call for an efficient ultra-low rate code design. In this work, we propose a novel coding scheme tailored for PUFs based on Polar codes and a low-complexity version of automorphism ensemble decoding (AED). Notably, our serial AED scheme reuses a single successive cancellation (SC) decoder across multiple decoding attempts. By introducing cascaded and recursive interleavers, we efficiently scale the number of AED candidates without requiring expensive large multiplexers. An aggressive quantization strategy of only 3 bits per message further reduces the area requirements of the underlying SC decoder. The resulting coding scheme achieves the same block error rate of $10^{-6}$ as our baseline based on Bose-Ray-Chaudhuri-Hocquenghem (BCH) codes while requiring 1.75x fewer codeword bits to encode the same K = 312 payload bits. This reduction translates directly into 1.75x less helper data storage and, consequently, a smaller overall chip area. - Open Access
Sionna Research Kit: A GPU-Accelerated Research Platform for AI-RAN
We introduce the NVIDIA Sionna Research Kit, a GPU-accelerated research platform for developing and testing AI/ML algorithms in 5G NR cellular networks. Powered by the NVIDIA Jetson AGX Orin, the platform leverages accelerated computing to deliver high throughput and real-time signal processing, while offering the flexibility of a software-defined stack. Built on OpenAirInterface (OAI), it unlocks a broad range of research opportunities. These include developing 5G NR and ORAN compliant algorithms, collecting real-world data for AI/ML training, and rapidly deploying innovative solutions in a very affordable testbed. Additionally, AI/ML hardware acceleration promotes the exploration of use cases in edge computing and AI radio access networks (AI-RAN). To demonstrate the capabilities, we deploy a real-time neural receiver - trained with NVIDIA Sionna and using the NVIDIA TensorRT library for inference - in a 5G NR cellular network using commercial user equipment. The code examples will be made publicly available, enabling researchers to adopt and extend the platform for their own projects.
Open AccessSionna Research Kit: A GPU-Accelerated Research Platform for AI-RAN
We introduce the NVIDIA Sionna Research Kit, a GPU-accelerated research platform for developing and testing AI/ML algorithms in 5G NR cellular networks. Powered by the NVIDIA Jetson AGX Orin, the platform leverages accelerated computing to deliver high throughput and real-time signal processing, while offering the flexibility of a software-defined stack. Built on OpenAirInterface (OAI), it unlocks a broad range of research opportunities. These include developing 5G NR and ORAN compliant algorithms, collecting real-world data for AI/ML training, and rapidly deploying innovative solutions in a very affordable testbed. Additionally, AI/ML hardware acceleration promotes the exploration of use cases in edge computing and AI radio access networks (AI-RAN). To demonstrate the capabilities, we deploy a real-time neural receiver - trained with NVIDIA Sionna and using the NVIDIA TensorRT library for inference - in a 5G NR cellular network using commercial user equipment. The code examples will be made publicly available, enabling researchers to adopt and extend the platform for their own projects.
- Top Cited PapersOpen Access
- 4 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Machine Learning in Communications and Networking
- Vol. 2, 1527-1539
Learning Radio Environments by Differentiable Ray Tracing
Ray tracing (RT) is instrumental in 6G research in order to generate spatially-consistent and environment-specific channel impulse responses (CIRs). While acquiring accurate scene geometries is now relatively straightforward, determining material characteristics requires precise calibration using channel measurements. We therefore introduce a novel gradient-based calibration method, complemented by differentiable parametrizations of material properties, scattering and antenna patterns. Our method seamlessly integrates with differentiable ray tracers that enable the computation of derivatives of CIRs with respect to these parameters. Essentially, we approach field computation as a large computational graph wherein parameters are trainable akin to weights of a neural network (NN). We have validated our method using both synthetic data and real-world indoor channel measurements, employing a distributed multiple-input multiple-output (MIMO) channel sounder.
- 4 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Machine Learning in Communications and Networking
- Vol. 2, 1527-1539
Top Cited PapersOpen AccessLearning Radio Environments by Differentiable Ray Tracing
Ray tracing (RT) is instrumental in 6G research in order to generate spatially-consistent and environment-specific channel impulse responses (CIRs). While acquiring accurate scene geometries is now relatively straightforward, determining material characteristics requires precise calibration using channel measurements. We therefore introduce a novel gradient-based calibration method, complemented by differentiable parametrizations of material properties, scattering and antenna patterns. Our method seamlessly integrates with differentiable ray tracers that enable the computation of derivatives of CIRs with respect to these parameters. Essentially, we approach field computation as a large computational graph wherein parameters are trainable akin to weights of a neural network (NN). We have validated our method using both synthetic data and real-world indoor channel measurements, employing a distributed multiple-input multiple-output (MIMO) channel sounder.
- Open Access
Design of a Standard-Compliant Real-Time Neural Receiver for 5G NR
We detail the steps required to deploy a multi-user multiple-input multiple-output (MU-MIMO) neural receiver (NRX) in an actual cellular communication system. This raises several exciting research challenges, including the need for real-time inference and compatibility with the 5G NR standard. As the network configuration in a practical setup can change dynamically within milliseconds, we propose an adaptive NRX architecture capable of supporting dynamic modulation and coding scheme (MCS) configurations without the need for any re-training and without additional inference cost. We optimize the latency of the neural network (NN) architecture to achieve inference times of less than 1ms on an NVIDIA A100 GPU using the TensorRT inference library. These latency constraints effectively limit the size of the NN and we quantify the resulting signal-to-noise ratio (SNR) degradation as less than 0.7 dB when compared to a preliminary non-real-time NRX architecture. Finally, we explore the potential for site-specific adaptation of the receiver by investigating the required size of the training dataset and the number of fine-tuning iterations to optimize the NRX for specific radio environments using a ray tracing-based channel model. The resulting NRX is ready for deployment in a real-time 5G NR system and the source code including the TensorRT experiments is available online.
Open AccessDesign of a Standard-Compliant Real-Time Neural Receiver for 5G NR
We detail the steps required to deploy a multi-user multiple-input multiple-output (MU-MIMO) neural receiver (NRX) in an actual cellular communication system. This raises several exciting research challenges, including the need for real-time inference and compatibility with the 5G NR standard. As the network configuration in a practical setup can change dynamically within milliseconds, we propose an adaptive NRX architecture capable of supporting dynamic modulation and coding scheme (MCS) configurations without the need for any re-training and without additional inference cost. We optimize the latency of the neural network (NN) architecture to achieve inference times of less than 1ms on an NVIDIA A100 GPU using the TensorRT inference library. These latency constraints effectively limit the size of the NN and we quantify the resulting signal-to-noise ratio (SNR) degradation as less than 0.7 dB when compared to a preliminary non-real-time NRX architecture. Finally, we explore the potential for site-specific adaptation of the receiver by investigating the required size of the training dataset and the number of fine-tuning iterations to optimize the NRX for specific radio environments using a ray tracing-based channel model. The resulting NRX is ready for deployment in a real-time 5G NR system and the source code including the TensorRT experiments is available online.
- Open Access
A Neural Receiver for 5G NR Multi-user MIMO
We introduce a neural network (NN)-based multiuser multiple-input multiple-output (MU-MIMO) receiver with 5G New Radio (5G NR) physical uplink shared channel (PUSCH) compatibility. The NN architecture is based on convolution layers to exploit the time and frequency correlation of the channel and a graph neural network (GNN) to handle multiple users. The proposed architecture adapts to an arbitrary number of sub-carriers and supports a varying number of multiple-input multiple-output (MIMO) layers and users without the need for any retraining. The receiver operates on an entire 5G NR slot, i.e., processes the entire received orthogonal frequency division multiplexing (OFDM) time-frequency resource grid by jointly performing channel estimation, equalization, and demapping. The proposed architecture operates less than 1 dB away from a baseline using linear minimum mean square error (LMMSE) channel estimation with K-best detection but benefits from a significantly lower computational complexity. We show the importance of a carefully designed training process such that the trained receiver is universal for a wide range of different unseen channel conditions. Finally, we demonstrate the results of a hardware-in-the-loop verification based on 3GPP compliant conformance test scenarios.
Open AccessA Neural Receiver for 5G NR Multi-user MIMO
We introduce a neural network (NN)-based multiuser multiple-input multiple-output (MU-MIMO) receiver with 5G New Radio (5G NR) physical uplink shared channel (PUSCH) compatibility. The NN architecture is based on convolution layers to exploit the time and frequency correlation of the channel and a graph neural network (GNN) to handle multiple users. The proposed architecture adapts to an arbitrary number of sub-carriers and supports a varying number of multiple-input multiple-output (MIMO) layers and users without the need for any retraining. The receiver operates on an entire 5G NR slot, i.e., processes the entire received orthogonal frequency division multiplexing (OFDM) time-frequency resource grid by jointly performing channel estimation, equalization, and demapping. The proposed architecture operates less than 1 dB away from a baseline using linear minimum mean square error (LMMSE) channel estimation with K-best detection but benefits from a significantly lower computational complexity. We show the importance of a carefully designed training process such that the trained receiver is universal for a wide range of different unseen channel conditions. Finally, we demonstrate the results of a hardware-in-the-loop verification based on 3GPP compliant conformance test scenarios.
- 12 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Communications
- Vol. 71 (2) , 837-850
Learning Joint Detection, Equalization and Decoding for Short-Packet Communications
We propose and practically demonstrate a joint detection and decoding scheme for short-packet wireless communications in scenarios that require to first detect the presence of a message before actually decoding it. For this, we extend the recently proposed serial Turbo-autoencoder neural network (NN) architecture and train it to find short messages that can be, all “at once”, detected, synchronized, equalized and decoded when sent over an unsynchronized channel with memory. The conceptional advantage of the proposed system stems from a holistic message structure with superimposed pilots for joint detection and decoding without the need of relying on a dedicated preamble. This results not only in a higher spectral efficiency, but also translates into the possibility of shorter messages compared to using a dedicated preamble. We compare the detection error rate (DER), bit error rate (BER) and block error rate (BLER) performance of the proposed system with a hand-crafted state-of-the-art conventional baseline and our simulations show a significant advantage of the proposed autoencoder-based system over the conventional baseline in every scenario up to messages conveying k =96 information bits. Finally, we practically evaluate and confirm the improved performance of the proposed system over-the-air (OTA) using a software-defined radio (SDR)-based measurement testbed.
- 12 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Communications
- Vol. 71 (2) , 837-850
Learning Joint Detection, Equalization and Decoding for Short-Packet Communications
We propose and practically demonstrate a joint detection and decoding scheme for short-packet wireless communications in scenarios that require to first detect the presence of a message before actually decoding it. For this, we extend the recently proposed serial Turbo-autoencoder neural network (NN) architecture and train it to find short messages that can be, all “at once”, detected, synchronized, equalized and decoded when sent over an unsynchronized channel with memory. The conceptional advantage of the proposed system stems from a holistic message structure with superimposed pilots for joint detection and decoding without the need of relying on a dedicated preamble. This results not only in a higher spectral efficiency, but also translates into the possibility of shorter messages compared to using a dedicated preamble. We compare the detection error rate (DER), bit error rate (BER) and block error rate (BLER) performance of the proposed system with a hand-crafted state-of-the-art conventional baseline and our simulations show a significant advantage of the proposed autoencoder-based system over the conventional baseline in every scenario up to messages conveying k =96 information bits. Finally, we practically evaluate and confirm the improved performance of the proposed system over-the-air (OTA) using a software-defined radio (SDR)-based measurement testbed.
- Open Access
Graph Neural Networks for Channel Decoding
In this work, we propose a fully differentiable graph neural network (GNN)-based architecture for channel decoding and showcase a competitive decoding performance for various coding schemes, such as low-density parity-check (LDPC) and BCH codes. The idea is to let a neural network (NN) learn a generalized message passing algorithm over a given graph that represents the forward error correction (FEC) code structure by replacing node and edge message updates with trainable functions. Contrary to many other deep learning-based decoding approaches, the proposed solution enjoys scalability to arbitrary block lengths and the training is not limited by the curse of dimensionality. We benchmark our proposed decoder against state-of-the-art in conventional channel decoding as well as against recent deep learning-based results. For the (63,45) BCH code, our solution outperforms weighted belief propagation (BP) decoding by approximately 0.4 dB with significantly less decoding iterations and even for 5G NR LDPC codes, we observe a competitive performance when compared to conventional BP decoding. For the BCH codes, the resulting GNN decoder can be fully parametrized with only 9640 weights.
Open AccessGraph Neural Networks for Channel Decoding
In this work, we propose a fully differentiable graph neural network (GNN)-based architecture for channel decoding and showcase a competitive decoding performance for various coding schemes, such as low-density parity-check (LDPC) and BCH codes. The idea is to let a neural network (NN) learn a generalized message passing algorithm over a given graph that represents the forward error correction (FEC) code structure by replacing node and edge message updates with trainable functions. Contrary to many other deep learning-based decoding approaches, the proposed solution enjoys scalability to arbitrary block lengths and the training is not limited by the curse of dimensionality. We benchmark our proposed decoder against state-of-the-art in conventional channel decoding as well as against recent deep learning-based results. For the (63,45) BCH code, our solution outperforms weighted belief propagation (BP) decoding by approximately 0.4 dB with significantly less decoding iterations and even for 5G NR LDPC codes, we observe a competitive performance when compared to conventional BP decoding. For the BCH codes, the resulting GNN decoder can be fully parametrized with only 9640 weights.
- Open Access
Adaptive Neural Network-based OFDM Receivers
We propose and examine the idea of continuously adapting state-of-the-art neural network (NN)-based orthogonal frequency division multiplex (OFDM) receivers to current channel conditions. This online adaptation via retraining is mainly motivated by two reasons: First, receiver design typically focuses on the universal optimal performance for a wide range of possible channel realizations. However, in actual applications and within short time intervals, only a subset of these channel parameters is likely to occur, as macro parameters, e.g., the maximum channel delay, can assumed to be static. Second, in-the-field alterations like temporal interferences or other conditions out of the originally intended specifications can occur on a practical (real-world) transmission. While conventional (filter-based) systems would require reconfiguration or additional signal processing to cope with these unforeseen conditions, NN-based receivers can learn to mitigate previously unseen effects even after their deployment. For this, we showcase on-the-fly adaption to current channel conditions and temporal alterations solely based on recovered labels from an outer forward error correction (FEC) code without any additional piloting overhead. To underline the flexibility of the proposed adaptive training, we showcase substantial gains for scenarios with static channel macro parameters, for out-of-specification usage and for interference compensation.
Open AccessAdaptive Neural Network-based OFDM Receivers
We propose and examine the idea of continuously adapting state-of-the-art neural network (NN)-based orthogonal frequency division multiplex (OFDM) receivers to current channel conditions. This online adaptation via retraining is mainly motivated by two reasons: First, receiver design typically focuses on the universal optimal performance for a wide range of possible channel realizations. However, in actual applications and within short time intervals, only a subset of these channel parameters is likely to occur, as macro parameters, e.g., the maximum channel delay, can assumed to be static. Second, in-the-field alterations like temporal interferences or other conditions out of the originally intended specifications can occur on a practical (real-world) transmission. While conventional (filter-based) systems would require reconfiguration or additional signal processing to cope with these unforeseen conditions, NN-based receivers can learn to mitigate previously unseen effects even after their deployment. For this, we showcase on-the-fly adaption to current channel conditions and temporal alterations solely based on recovered labels from an outer forward error correction (FEC) code without any additional piloting overhead. To underline the flexibility of the proposed adaptive training, we showcase substantial gains for scenarios with static channel macro parameters, for out-of-specification usage and for interference compensation.
- Open Access
Serial vs. Parallel Turbo-Autoencoders and Accelerated Training for Learned Channel Codes
Attracted by its scalability towards practical codeword lengths, we revisit the idea of Turbo-autoencoders for end-to-end learning of PHY-Layer communications. For this, we study the existing concepts of Turbo-autoencoders from the literature and compare the concept with state-of-the-art classical coding schemes. We propose a new component-wise training algorithm based on the idea of Gaussian a priori distributions that reduces the overall training time by almost a magnitude. Further, we propose a new serial architecture inspired by classical serially concatenated Turbo code structures and show that a carefully optimized interface between the two component autoencoders is required. To the best of our knowledge, these serial Turbo autoencoder structures are the best known neural network based learned sequences that can be trained from scratch without any required expert knowledge in the domain of channel codes.
Open AccessSerial vs. Parallel Turbo-Autoencoders and Accelerated Training for Learned Channel Codes
Attracted by its scalability towards practical codeword lengths, we revisit the idea of Turbo-autoencoders for end-to-end learning of PHY-Layer communications. For this, we study the existing concepts of Turbo-autoencoders from the literature and compare the concept with state-of-the-art classical coding schemes. We propose a new component-wise training algorithm based on the idea of Gaussian a priori distributions that reduces the overall training time by almost a magnitude. Further, we propose a new serial architecture inspired by classical serially concatenated Turbo code structures and show that a carefully optimized interface between the two component autoencoders is required. To the best of our knowledge, these serial Turbo autoencoder structures are the best known neural network based learned sequences that can be trained from scratch without any required expert knowledge in the domain of channel codes.