Cornelius Hellge
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Cornelius Hellge has a total of 36 co-authors. Here are the top co-authors by number of shared publications:
- Thomas Schierl
Fraunhofer HHI Berlin Germany
7 shared publications Thomas Schierl
Video Communication and Applications Department, Fraunhofer Heinrich Hertz Institute, Berlin, Germany
5 shared publications- Barış Göktepe
Video Communication and Applications Department, Fraunhofer Heinrich Hertz Institute, Berlin, Germany
4 shared publications - Jangwoo Son
Fraunhofer Heinrich Hertz Institute, Berlin, Germany
2 shared publications Robert Skupin
Video Coding and Analytics Department, Multimedia Communications Group, Fraunhofer Heinrich-Hertz Institute, Berlin, Germany
2 shared publications- Serhan Gül
Fraunhofer HHI Berlin Germany
2 shared publications
Latest Publications
- Open Access
- 27 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 5, 841-849
Adaptable L4S Congestion Control for Cloud-Based Real-Time Streaming Over 5G
Achieving reliable low-latency streaming on real-time immersive services that require seamless interaction has been of increasing importance recently. To cope with such an immersive service requirement, IETF and 3GPP defined Low Latency, Low Loss, and Scalable Throughput (L4S) architecture and terminologies to enable delay-critical applications to achieve low congestion and scalable bitrate control over 5 G. With low-latency applications in mind, this paper presents a cloud-based streaming system using WebRTC for real-time communication with an adaptable L4S congestion control (aL4S-CC). aL4S-CC is designed to prevent the target service from surpassing a required end-to-end latency. It is evaluated against existing congestion controls GCC and ScreamV2 across two configurations: 1) standard L4S (sL4S) which has no knowledge of Explicit Congestion Notification (ECN) marking scheme information; 2) conscious L4S (cL4S) which recognizes the ECN marking scheme information. The results show that aL4S-CC achieves high link utilization with low latency while maintaining good performance in terms of fairness, and cL4S improves sL4S's performance by having an efficient trade-off between link utilization and latency. In the entire simulation, the gain of link utilization on cL4S is 1.4 %, 4 %, and 17.9 % on average compared to sL4S, GCC, and ScreamV2, respectively, and the ratio of duration exceeding the target queuing delay achieves the lowest values of 1 % and 0.9 % for cL4S and sL4S, respectively.
- 27 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Signal Processing
- Vol. 5, 841-849
Open AccessAdaptable L4S Congestion Control for Cloud-Based Real-Time Streaming Over 5G
Achieving reliable low-latency streaming on real-time immersive services that require seamless interaction has been of increasing importance recently. To cope with such an immersive service requirement, IETF and 3GPP defined Low Latency, Low Loss, and Scalable Throughput (L4S) architecture and terminologies to enable delay-critical applications to achieve low congestion and scalable bitrate control over 5 G. With low-latency applications in mind, this paper presents a cloud-based streaming system using WebRTC for real-time communication with an adaptable L4S congestion control (aL4S-CC). aL4S-CC is designed to prevent the target service from surpassing a required end-to-end latency. It is evaluated against existing congestion controls GCC and ScreamV2 across two configurations: 1) standard L4S (sL4S) which has no knowledge of Explicit Congestion Notification (ECN) marking scheme information; 2) conscious L4S (cL4S) which recognizes the ECN marking scheme information. The results show that aL4S-CC achieves high link utilization with low latency while maintaining good performance in terms of fairness, and cL4S improves sL4S's performance by having an efficient trade-off between link utilization and latency. In the entire simulation, the gain of link utilization on cL4S is 1.4 %, 4 %, and 17.9 % on average compared to sL4S, GCC, and ScreamV2, respectively, and the ratio of duration exceeding the target queuing delay achieves the lowest values of 1 % and 0.9 % for cL4S and sL4S, respectively.
- 15 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 31 (10709908) , 1504-1508
Improving QoE-Privacy Tradeoff in XR Streaming
Viewpoint and position (VP)-adaptive mixed reality (XR) streaming requires uploading the trajectory of user behavior, thereby causing privacy leakage. Preserving privacy leads to the performance loss of VP prediction of users, subsequently degrading the quality of experience (QoE). This work is the first to improve the QoE-privacy tradeoff for XR streaming. We find the key difference of sample importance for privacy attack and VP prediction, based on which we propose a framework to improve the tradeoff. By designing the noisy entropy function and remapping function, it can achieve a better tradeoff than simply adding the noise to the actual trajectory. The performance of the framework is evaluated with the state-of-the-art VP predictors and practical XR streaming platforms. The results show that the loss of QoE can be mitigated by 55% $\sim$ 100% to achieve the same privacy level as simply adding noise. When the privacy level achieves the maximum, the QoE is degraded by 0 $\sim$ 3%, compared to VP-adaptive streaming without any privacy-preserving.
- 15 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Letters
- Vol. 31 (10709908) , 1504-1508
Improving QoE-Privacy Tradeoff in XR Streaming
Viewpoint and position (VP)-adaptive mixed reality (XR) streaming requires uploading the trajectory of user behavior, thereby causing privacy leakage. Preserving privacy leads to the performance loss of VP prediction of users, subsequently degrading the quality of experience (QoE). This work is the first to improve the QoE-privacy tradeoff for XR streaming. We find the key difference of sample importance for privacy attack and VP prediction, based on which we propose a framework to improve the tradeoff. By designing the noisy entropy function and remapping function, it can achieve a better tradeoff than simply adding the noise to the actual trajectory. The performance of the framework is evaluated with the state-of-the-art VP predictors and practical XR streaming platforms. The results show that the loss of QoE can be mitigated by 55% $\sim$ 100% to achieve the same privacy level as simply adding noise. When the privacy level achieves the maximum, the QoE is degraded by 0 $\sim$ 3%, compared to VP-adaptive streaming without any privacy-preserving.
- Open Access
- 17 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 23 (1) , 31-44
Distributed Machine-Learning for Early HARQ Feedback Prediction in Cloud RANs
In this work, we propose novel HARQ prediction schemes for Cloud RANs (C-RANs) that use feedback over a rate-limited feedback channel (2 - 6 bits) from the Remote Radio Heads (RRHs) to predict at the User Equipment (UE) the decoding outcome at the BaseBand Unit (BBU) ahead of actual decoding. In particular, we propose a Dual Autoencoding 2-Stage Gaussian Mixture Model (DA2SGMM) that is trained in an end-to-end fashion over the whole C-RAN setup. Using realistic link-level simulations in the sub-THz band at 100 GHz, we show that the novel DA2SGMM HARQ prediction scheme clearly outperforms all other adapted and state-of-the-art schemes. The DA2SGMM shows a superior performance in terms of blockage detection as well as HARQ prediction in the no-blockage and single-blockage cases. In particular, the DA2SGMM with 4 bit feedback achieves a more than 200 % higher throughput in average compared to its best alternative. Compared to regular HARQ, the DA2SGMM reduces the maximum transmission latency by more than 72.4 %, while maintaining more than 75 % of the throughput in the no-blockage scenario. In the single-blockage scenario, DA2SGMM significantly increases the throughput for most of the evaluated Signal-to-Noise-Ratios (SNRs) compared to regular HARQ.
- 17 May 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Wireless Communications
- Vol. 23 (1) , 31-44
Open AccessDistributed Machine-Learning for Early HARQ Feedback Prediction in Cloud RANs
In this work, we propose novel HARQ prediction schemes for Cloud RANs (C-RANs) that use feedback over a rate-limited feedback channel (2 - 6 bits) from the Remote Radio Heads (RRHs) to predict at the User Equipment (UE) the decoding outcome at the BaseBand Unit (BBU) ahead of actual decoding. In particular, we propose a Dual Autoencoding 2-Stage Gaussian Mixture Model (DA2SGMM) that is trained in an end-to-end fashion over the whole C-RAN setup. Using realistic link-level simulations in the sub-THz band at 100 GHz, we show that the novel DA2SGMM HARQ prediction scheme clearly outperforms all other adapted and state-of-the-art schemes. The DA2SGMM shows a superior performance in terms of blockage detection as well as HARQ prediction in the no-blockage and single-blockage cases. In particular, the DA2SGMM with 4 bit feedback achieves a more than 200 % higher throughput in average compared to its best alternative. Compared to regular HARQ, the DA2SGMM reduces the maximum transmission latency by more than 72.4 %, while maintaining more than 75 % of the throughput in the no-blockage scenario. In the single-blockage scenario, DA2SGMM significantly increases the throughput for most of the evaluated Signal-to-Noise-Ratios (SNRs) compared to regular HARQ.
- Open Access
- 31 August 2021
- Published by MDPI AG in Mathematics
- Vol. 9 (17) , 2104
Analytical Model and Feedback Predictor Optimization for Combined Early-HARQ and HARQ
In order to fulfill the stringent Ultra-Reliable Low Latency Communication (URLLC) requirements towards Fifth Generation (5G) mobile networks, early-Hybrid Automatic Repeat reQuest (e-HARQ) schemes have been introduced, aimed at providing faster feedback and thus earlier retransmission. The performance of e-HARQ prediction strongly depends on the classification mechanism, data length, threshold value. In this paper, we propose an analytical model that incorporates e-HARQ and Hybrid Automatic Repeat reQuest (HARQ) functionalities in terms of two phases in discrete time. The model implies a fast and accurate way to get the main performance measures, and apply optimization analysis to find the optimal values used in predictor’s classification. We employ realistic data for transition probabilities obtained by means of 5G link-level simulations and conduct extensive experimental analysis. The results show that at false positive probability of , the e-HARQ prediction with the found optimal parameters can achieve around 20% of gain over HARQ at False Negative (FN) of and around 7.5% at FN of in terms of a mean spending time before successful delivery.- 31 August 2021
- Published by MDPI AG in Mathematics
- Vol. 9 (17) , 2104
Open AccessAnalytical Model and Feedback Predictor Optimization for Combined Early-HARQ and HARQ
In order to fulfill the stringent Ultra-Reliable Low Latency Communication (URLLC) requirements towards Fifth Generation (5G) mobile networks, early-Hybrid Automatic Repeat reQuest (e-HARQ) schemes have been introduced, aimed at providing faster feedback and thus earlier retransmission. The performance of e-HARQ prediction strongly depends on the classification mechanism, data length, threshold value. In this paper, we propose an analytical model that incorporates e-HARQ and Hybrid Automatic Repeat reQuest (HARQ) functionalities in terms of two phases in discrete time. The model implies a fast and accurate way to get the main performance measures, and apply optimization analysis to find the optimal values used in predictor’s classification. We employ realistic data for transition probabilities obtained by means of 5G link-level simulations and conduct extensive experimental analysis. The results show that at false positive probability of , the e-HARQ prediction with the found optimal parameters can achieve around 20% of gain over HARQ at False Negative (FN) of and around 7.5% at FN of in terms of a mean spending time before successful delivery. - 22 December 2020
- Published by Springer Nature
- p. 222-239
Analytical Model of Early HARQ Feedback Prediction
- 22 December 2020
- Published by Springer Nature
- p. 222-239
Analytical Model of Early HARQ Feedback Prediction
- Open Access
- 12 October 2020
- Published by Association for Computing Machinery (ACM)
- p. 3632-3641
Kalman Filter-based Head Motion Prediction for Cloud-based Mixed Reality
Volumetric video allows viewers to experience highly-realistic 3D content with six degrees of freedom in mixed reality (MR) environments. Rendering complex volumetric videos can require a prohibitively high amount of computational power for mobile devices. A promising technique to reduce the computational burden on mobile devices is to perform the rendering at a cloud server. However, cloud-based rendering systems suffer from an increased interaction (motion-to-photon) latency that may cause registration errors in MR environments. One way of reducing the effective latency is to predict the viewer's head pose and render the corresponding view from the volumetric video in advance. In this paper, we design a Kalman filter for head motion prediction in our cloud-based volumetric video streaming system. We analyze the performance of our approach using recorded head motion traces and compare its performance to an autoregression model for different prediction intervals (look-ahead times). Our results show that the Kalman filter can predict head orientations 0.5 degrees more accurately than the autoregression model for a look-ahead time of 60 ms.
- 12 October 2020
- Published by Association for Computing Machinery (ACM)
- p. 3632-3641
Open AccessKalman Filter-based Head Motion Prediction for Cloud-based Mixed Reality
Volumetric video allows viewers to experience highly-realistic 3D content with six degrees of freedom in mixed reality (MR) environments. Rendering complex volumetric videos can require a prohibitively high amount of computational power for mobile devices. A promising technique to reduce the computational burden on mobile devices is to perform the rendering at a cloud server. However, cloud-based rendering systems suffer from an increased interaction (motion-to-photon) latency that may cause registration errors in MR environments. One way of reducing the effective latency is to predict the viewer's head pose and render the corresponding view from the volumetric video in advance. In this paper, we design a Kalman filter for head motion prediction in our cloud-based volumetric video streaming system. We analyze the performance of our approach using recorded head motion traces and compare its performance to an autoregression model for different prediction intervals (look-ahead times). Our results show that the Kalman filter can predict head orientations 0.5 degrees more accurately than the autoregression model for a look-ahead time of 60 ms.
- Open Access
Low-latency Cloud-based Volumetric Video Streaming Using Head Motion Prediction
Volumetric video is an emerging key technology for immersive representation of 3D spaces and objects. Rendering volumetric video requires lots of computational power which is challenging especially for mobile devices. To mitigate this, we developed a streaming system that renders a 2D view from the volumetric video at a cloud server and streams a 2D video stream to the client. However, such network-based processing increases the motion-to-photon (M2P) latency due to the additional network and processing delays. In order to compensate the added latency, prediction of the future user pose is necessary. We developed a head motion prediction model and investigated its potential to reduce the M2P latency for different look-ahead times. Our results show that the presented model reduces the rendering errors caused by the M2P latency compared to a baseline system in which no prediction is performed.
Open AccessLow-latency Cloud-based Volumetric Video Streaming Using Head Motion Prediction
Volumetric video is an emerging key technology for immersive representation of 3D spaces and objects. Rendering volumetric video requires lots of computational power which is challenging especially for mobile devices. To mitigate this, we developed a streaming system that renders a 2D view from the volumetric video at a cloud server and streams a 2D video stream to the client. However, such network-based processing increases the motion-to-photon (M2P) latency due to the additional network and processing delays. In order to compensate the added latency, prediction of the future user pose is necessary. We developed a head motion prediction model and investigated its potential to reduce the M2P latency for different look-ahead times. Our results show that the presented model reduces the rendering errors caused by the M2P latency compared to a baseline system in which no prediction is performed.
- 30 April 2020
- Published by Springer Nature
- p. 777-780
A Virtualized Video Surveillance System for Public Transportation
- 30 April 2020
- Published by Springer Nature
- p. 777-780
A Virtualized Video Surveillance System for Public Transportation
- Open Access
- 8 August 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Selected Areas in Communications
- Vol. 37 (11) , 2573-2587
Enhanced Machine Learning Techniques for Early HARQ Feedback Prediction in 5G
We investigate Early Hybrid Automatic Repeat re-Quest (E-HARQ) feedback schemes enhanced by machine learning techniques as a path towards ultra-reliable and low-latency communication (URLLC). To this end, we propose machine learning methods to predict the outcome of the decoding process ahead of the end of the transmission. We discuss different input features and classification algorithms ranging from traditional methods to newly developed supervised autoencoders. These methods are evaluated based on their prospects of complying with the URLLC requirements of effective block error rates below 10-5 at small latency overheads. We provide realistic performance estimates in a system model incorporating scheduling effects to demonstrate the feasibility of E-HARQ across different signal-to-noise ratios, subcode lengths, channel conditions and system loads, and show the benefit over regular HARQ and existing E-HARQ schemes without machine learning.
- 8 August 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Selected Areas in Communications
- Vol. 37 (11) , 2573-2587
Open AccessEnhanced Machine Learning Techniques for Early HARQ Feedback Prediction in 5G
We investigate Early Hybrid Automatic Repeat re-Quest (E-HARQ) feedback schemes enhanced by machine learning techniques as a path towards ultra-reliable and low-latency communication (URLLC). To this end, we propose machine learning methods to predict the outcome of the decoding process ahead of the end of the transmission. We discuss different input features and classification algorithms ranging from traditional methods to newly developed supervised autoencoders. These methods are evaluated based on their prospects of complying with the URLLC requirements of effective block error rates below 10-5 at small latency overheads. We provide realistic performance estimates in a system model incorporating scheduling effects to demonstrate the feasibility of E-HARQ across different signal-to-noise ratios, subcode lengths, channel conditions and system loads, and show the benefit over regular HARQ and existing E-HARQ schemes without machine learning.
- Open Access
- 14 February 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Emerging and Selected Topics in Circuits and Systems
- Vol. 9 (1) , 18-28
Delay Impact on MPEG OMAF’s Tile-Based Viewport-Dependent 360° Video Streaming
With the continuously increasing attention that 360° video streaming is drawing, several solutions have been developed lately. They can be grouped into two main categories, with the first one being the viewport-independent category that consists of encoding the whole 360° video content using a particular projection, e.g., Equirectangular Projection or Cubemap Projection, without taking any viewport orientation into account. Such approaches waste resources since content not being watched by the user is encoded with the same visual fidelity as the content actually watched. The second category, the viewport-dependent category, relies on techniques that allow for viewport adaptivity. One approach is to apply viewport-dependent projections, wherein the spherical 360° video is mapped onto a rectangular frame in such a way that a specific viewport is mapped to a comparatively larger part of the rectangular frame than the rest of the content. Another approach, as specified in MPEG OMAF, is to use tile-based streaming, with the 360° video being offered in a tiled manner at various resolutions. Thus, each user can retrieve the tiles at different resolutions so that the high-resolution tiles match its viewport. Although viewport-dependent approaches allow providing better visual quality at the viewport, the end-toend delay is critical, since it has an impact on the time needed by clients to adapt to user movements so that movements are reflected on the retrieved content. In this paper, we analyze the impact of the delay and introduce various means to reduce the impact on the observed fidelity.
- 14 February 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal on Emerging and Selected Topics in Circuits and Systems
- Vol. 9 (1) , 18-28
Open AccessDelay Impact on MPEG OMAF’s Tile-Based Viewport-Dependent 360° Video Streaming
With the continuously increasing attention that 360° video streaming is drawing, several solutions have been developed lately. They can be grouped into two main categories, with the first one being the viewport-independent category that consists of encoding the whole 360° video content using a particular projection, e.g., Equirectangular Projection or Cubemap Projection, without taking any viewport orientation into account. Such approaches waste resources since content not being watched by the user is encoded with the same visual fidelity as the content actually watched. The second category, the viewport-dependent category, relies on techniques that allow for viewport adaptivity. One approach is to apply viewport-dependent projections, wherein the spherical 360° video is mapped onto a rectangular frame in such a way that a specific viewport is mapped to a comparatively larger part of the rectangular frame than the rest of the content. Another approach, as specified in MPEG OMAF, is to use tile-based streaming, with the 360° video being offered in a tiled manner at various resolutions. Thus, each user can retrieve the tiles at different resolutions so that the high-resolution tiles match its viewport. Although viewport-dependent approaches allow providing better visual quality at the viewport, the end-toend delay is critical, since it has an impact on the time needed by clients to adapt to user movements so that movements are reflected on the retrieved content. In this paper, we analyze the impact of the delay and introduce various means to reduce the impact on the observed fidelity.