Deliverables
Deliverable 1.1: Internal web-based platform for collaboration
The hi-STAR project addresses one of the most critical challenges for the next generation wireless networks, which is integration of non-terrestrial networks with terrestrial 5G network. The general objective of the project is to develop flexible framework for integrated terrestrial 5G and Low-Earth-Orbit (LEO) satellite networks, where traffic management is performed with assistance of newly developed artificial intelligence methods. The project comprises seven work packages, and it is important to have a secured platform for internal document sharing among the project team members. This deliverable is a result of the work done in WP1 Subactivity T1.1 – Coordination. Deliverable D1.1 presents description of internal web-based platform for collaboration.
Deliverable 1.3: Quarterly progress report - Q1
During the first quarter of the project, the appropriate channel models are identified for various propagation scenarios. Typical fading channel models are applied to describe short-term fading episodes in microwave terrestrial links, and a more complex two-way model with diffuse power will be used to model the terrestrial millimeter-wave links. Shadowed Ricean and gamma-shadowed Ricean fading models are identified as appropriate for the satellite links. A method for reliability analysis of satellite links subjected to deep fades is proposed. Besides, the hi-STAR project team has managed to find several suitable software solutions for the physical layer implementation of DVB-S2x. The found software solutions were analyzed with particular reference to future development (the simplicity of implementing the solution on an embedded platform), code reusability, and performance.
Deliverable 1.3: Quarterly progress report - Q2
During the second quarter of the project, our focus was on the creation of an adequate simulation environment that describes the propagation in satellite and 5G communication links. Three papers are presented at the international conference IcETRAN, and one of these papers is selected as the best paper presented in the Section of Telecommunications. Also, one paper is presented at the international conference ICEST, and two papers are finally accepted for presentation at the conference CSNDSP, which will be held in Porto in July this year. A detailed analysis of the GNU radio SDR framework is completed, and the SW architecture for DVB-S2x physical layer implementation is defined.
Deliverable 1.3: Quarterly progress report - Q3
In the third quarter, our focus was on algorithms that could increase the reliability of low-latency links, HDL implementation of basic modem functions, and identifying handover algorithms that use a small number of attributes. We have proposed the adaptation method based on the genetic optimization algorithm that is incorporated into a state-of-the-art decoding algorithm, resulting in a low-complexity LDPC decoder with superior performance. Also, we have optimized DVB-S2X physical layer software implementation to achieve better usage of processing resources, and we have almost completed HDL implementation of the most critical blocks for DVB-S2x hw/sw codesign: LDPC and BCH decoders. Two papers are presented at the international conference CSNDSP 2022, which was held in Porto in July this year, and one paper is finally accepted for publication in the journal IEEE Communications Letters.
Deliverable 1.3: Quarterly progress report - Q4
In the fourth quarter, our focus was on developing more realistic channel simulators and hardware implementation of the error correction decoders. The simulator of the satellite-terrestrial link is improved and we evaluated the statistical parameters of the signal envelope over the terrestrial mm-Wave channel. HDL design of the standalone BCH decoder is completed while LDPC is still under development. The focus of this activity besides hardware implementation is on porting 5G NR chain to the Xilinx development platform. We continued work related to examining handover techniques based on the MADM and reinforced learning applicable to hybrid radio network access. This quarter we were mainly focused on the LEO/GEO satellite networks and we discussed the potential of doing horizontal handover within the satellite access networks. Further analysis steps were conducted regarding the choice between OpenAirInterface 5G-core network and free5GC solutions that have been selected and initially analyzed in the research done in the previous quarter (OpenAirInterface will be in focus regarding the 5G network core, but the free5GC user element emulation will be used for initial tests). Also, gateway design and gateway connection to considered 5G network core solutions have been analyzed.
Deliverable 1.3: Quarterly progress report - Q5
In the fifth quarter, our focus was on hybrid user terminal development. More realistic channel simulators that will be used in HUT are developed. We continued the implementation of missing hardware blocks and started the implementation of software infrastructure on the RF-SoC platform. The simulation environment applicable to model the radio access network (RAN) is created, and it consists of multiple LEO satellites with desired constellations. We started the research that will explain our findings related to multiple-RAN-connections user terminals and their ability to increase information throughput, reduce the number of handovers and minimize link blockage probability. The analysis of HUT and NC interconnection is performed, because this aspect has a great impact on the final PoC demo, as well as on HUT implementation.
Deliverable 1.3: Quarterly progress report - Q6
In the sixth quarter, our focus was on WP2, WP3, and WP4. In the reporting period, we have created and tested the simulation environment that incorporates the shadowed-Rician propagation model, for the satellite communication channels and a two-wave with diffuse-power channel model for the terrestrial communication. Also, we have proposed and tested an initial TCM (traffic control module) and verified the improvements in terms of achievable spectral efficiency and data loss rate. LDPC and BCH encoder and decoder accelerators for SATCOM are completed, and these accelerators support all standard codes from the DVB-S2X communication standard. Communication between RF-SoC boards and one PC host using an Ethernet interface is established using Linux OS. The initial version of the GUI application is developed to enable easier communication control.
Deliverable 1.3: Quarterly progress report - Q7
In the seventh quarter, our focus was on WP2, WP3, and WP4. In the reporting period, the 5G NR QAM mapper and demapper were implemented and tested on an RF-SoC platform. The integrated SD-FEC decoder has been wrapped into the functional 5G decoder IP core. For encoding, it was decided to use the soft 5G LDPC encoder that was previously developed to avoid interfacing with SD-FEC, as its resource consumption will not cause bottlenecks. Components were successfully connected into the functional 5G loop-back chain with a random data generator acting as a data source, 5G LDPC encoder, QAM mapper, AWGN noise generator, QAM demapper, LDPC decoder, and data checker that evaluates the decoding performance. Also, we proposed a generalization of the recently published adaptive diversity gradient-descent bit flipping (AD-GDBF) decoder, that incorporates several improvements that make it eligible for the additive white Gaussian channel and decoding of arbitrary linear block code. The effectiveness of the proposed method is verified on short Bose–Chaudhuri–Hocquenghem (BCH) codes, which are used in the DVB-S2X protocol applied in the satellite modem. In addition, we have continued our work regarding the design of an intelligent traffic control unit (ITCU). Our ongoing research is directed toward improving the neural network-based design of predictors of channel state information.
Deliverable 1.3: Quarterly progress report - Q8
In the eighth quarter, our focus was on WP3, WP4, and WP5. In the reporting period, the project team has improved the developed software application to enable better control of data flow. Integration of developed IP blocks is in progress. During the reporting period, we tested and verified our handover execution method and compared our solution with state-of-the-art machine learning (ML) techniques that can be alternatively used for the prediction of the communication channel. We only considered a low-complexity ML solution, given that the handover method is placed in the end-user terminal, which prohibits the use of complex optimization methods. One of the major activities was related to completing deliverable D5.1 related to functional gateway design for use in the PoC demo. As well, a milestone regarding the functional 5G network core was reached. Based on detected open research problems regarding asymmetric links in multipath transport protocols we have defined scenarios for testing various combinations of link parameters such as latency, packet losses, and bandwidth. Milestones M4.2 and M5.1 were reached. We inspected two open source environments - OpenAirInterface and free5GC, both are working according to 5G standards and both can be used as a network core in PoC demo (part of the results regarding this milestone was published at the conference IcETRAN 2023 and the paper was awarded in Telecommunications section - Best young researcher's paper). Two conference papers were presented at important international conferences (one paper is an invited paper). Our paper, published in the Sensors journal, was selected as the best paper in the area of Telecommunications, published in an international scientific journal in the year 2022/2023.
Deliverable 1.3: Quarterly progress report - Q9
Regarding 5G/Sat transceivers, coding and decoding accelerators are implemented and tested. Other elements of transceivers will be implemented in the software. Integration of accelerator components is stalled and the current focus is on the implementation of complete functional HUT based on Linux application. Developed accelerators will be eventually used in the later stages after the entire HUT communication chain is established. The work regarding task 4.3 was conducted in order to develop the management of data traffic steering and switching logic on the RFSoC platform. The software was upgraded to accommodate multiple parallel TCP/UDP servers and client connections, which are responsible for efficient data traffic steering. Moreover, the existing software's data control logic was expanded to enable the periodic transmission of channel statistics to the main gateway machine, where the core logic for the HUT is situated. To facilitate traffic switching, the HUT logic core implemented on the gateway machine in the form of a GUI application was extended to support the reception and analysis of data statistics. Work on test environment design to measure the impact of latency and packet loss rate for various combinations of their values to optimally define multipath transport protocol setting for different parameter values. Performed throughput measurements for different link parameter combinations for multipath transmission control protocol (MP-TCP). In order to examine the performance of HUT, channel conditions in terrestrial and land-satellite channels need to be emulated. Channel models are described by probability density functions and autocorrelation functions and both of these functions need to be appropriately emulated. For the land-satellite communication, a shadowed-Rician channel model is used, which is composed of scattering and shadowing components - independently modeled by Nakagami-m and Rayleigh distributions. Nakagami-m and Rayleigh distributed processes can be emulated by a modified Jakes model, or in a twostep approach 1) first model uncorrelated Nakagami-m and Rayleigh samples and then 2) use autoregressive moving average (ARMA) model to insert desired level of correlation among them. The second modeling approach is chosen due to easier implementation: it requires CORDIC blocks for sample generation and FIR and IIR filters for ARMA block. The terrestrial channel model is called TWDP (two-wave with diffuse power) fading model and it is emulated in similar fashion: first, uncorrelated samples are generated and then the ARMA block is used for final sample shaping. One paper was submitted to the international journal indexed in the JCR list. One paper was submitted in the international conference CSNDSP 2024.
Deliverable 1.3: Quarterly progress report - Q10
In the tenth quarter, our focus was on WP4, WP5, and WP6. In this reporting period, we analyzed handover strategies in land-mobile satellite communications. Furthermore, our research was oriented toward integrating our model into a global framework, to prove its usefulness in practice. Major work was done on the preparation and finishing of Deliverable D5.2 as this was the main task for this period in Subactivity 5.2. Classification of business use cases based on industrial sectors has been done and presented to one telecommunication operator. Different models were analyzed for the channel emulator and a two-step approach will be used. The initial architecture of the channel emulator is proposed. One paper was published in the international journal from the JCR list, in the journal Entropy (published in May 2024). Two project team members presented the project at the Internal workshop forum of Wireless Communication, TU Wien, Vienna, Austria (April 2024). Three project team members presented an online lecture for engineers employed at Telekom Srbija, Belgrade, Serbia (June 2024).
Deliverable 1.3: Quarterly progress report - Q11
In this reporting period, we have proposed a novel strategy for adaptive coding and modulation employment in land mobile satellite networks. The proposed solution incorporates machine learning techniques to predict channel state information and subsequently increase the overall spectral efficiency of the network. Majority of efforts in the previous quarter were on completion of the HUT framework that can support steering of data streams through satellite or terrestrial paths. Simple quality of service mechanism and channel model are implemented to get initial performance results. HUT framework is currently able to track percentage of lost packets on both data channels and have simple hysteresis threshold logic for switching data stream from one to another channel. One paper was published in the international journal from the JCR list, in the journal Electronics (published in September 2024), two papers were presented at international conferences CSNDSP 2024 and IcETRAN 2024, and four papers were submitted to the international conference TELFOR 2024.
Deliverable 1.4: Annual progress report - Y1
During the first project year, we were focused on algorithms that could increase the reliability of low-latency links. We performed a reliability analysis of satellite links subjected to deep fades, and we have developed an accurate simulator of the terrestrial-satellite channel. We proposed a novel framework that can be applied to increase the reliability of earth or satellite links, based on the genetic optimization algorithm that is incorporated into the recently proposed Gradient Descent Bit-Flipping Decoding with Momentum (GDBF-w/M). The resulting low-complexity decoder outperforms all state-of-the-art probabilistic bit-flipping decoders, and additionally, it can be trained to perform beyond BP decoding. We have developed the architecture of the hybrid modem physical layer. GNU radio and Open Air Interface are chosen for DVB-S2x and 5G modems. Different architectures are analyzed for the DVB-S2x physical layer in GNU radio in order to optimally use the processing platform. Hardware accelerators are developed for the most demanding blocks in the physical layer. The BCH decoder is completed while the LDPC decoder is still under development. During the first year of the project, after a detailed examination of several state-of-the-art network simulators, we started to create our own hybrid radio access MATLAB simulator, with the ability to estimate improvements of the hybrid access in terms of reliability and achievable information rates. We tested several state-of-the-art MADM (multi-attribute decision-making) algorithms and quantified the gain achieved by using hybrid access. The first-year research goal was also the design analysis of the 5G network core and gateway for hybrid access. Research of open source 5G network core solutions was conducted and two platforms were found as potential candidates – Open Air Interface and Free5GC. Also, possible approaches for gateway design were investigated, and from the selected architectures, we will select the one that matches the best to the solutions found in WP2 and WP3 that cover hybrid user terminal design and overall system architecture. The project website is created with all project details and achievements and is continuously updated. Dissemination materials, i.e. project presentation, poster, and fact sheet are created for promoting the main ideas of the project. Professional and social networks profiles are created and managed (LinkedIn, Facebook). Defined KPIs are regularly monitored. Project results are published in journals (3 papers) and conferences (7 papers).
Deliverable 1.4: Annual progress report - Y2
During the second project year, we were focused on the implementation of the modems, and developing the handover algorithms. hi-STAR project team completed the required hardware accelerators supporting the satellite and terrestrial communication channels. We managed to complete the functional 5G communication chain loop-back in hardware. Software infrastructure enabling communication between RF-SoC boards and PC is developed using an Ethernet interface. The software application is optimized to support better control of data flows. Development of the remaining functionalities of software infrastructure is in progress. Integration of developed IP blocks into the functional system is in progress. We created the simulation environment applicable to model the radio access network (RAN) which consists of multiple LEO satellites, with desired constellations. We have developed and tested our TCM unit and verified the benefits of establishing multi-RAN connections and performing user-centric handover operations. The important responsibilities included improving ITCU by incorporating machine-learning techniques to provide targeted performance of the HUT module, measured in terms of increased spectral efficiency and desired level of data loss rates. We compared our solution with state-of-the-art machine learning (ML) techniques that can be alternatively used for the prediction of the communication channel. For the Proof of Concept (PoC) demo, Network Core (NC) and Hybrid User Terminal (HUT) interconnection have been analyzed. Special attention has been given to nFAPI interface and its messages since open source NCs that have been previously analyzed use this interface. Gateway enables Access Traffic Steering, Switching, and Splitting (ATSSS) support which is mandatory for HUT. Multipath support is necessary for ATSSS functions, thus, an analysis of multipath protocols has been done. Based on the analysis results, MultiPath Transmission Control Protocol (MP-TCP) is selected for initial PoC demo tests because it has been chosen as the main approach at the transport layer by 3GPP. We have defined multiple scenarios to test the impact of asymmetry between access links on the overall performance of ATSSS functions. The website of the project https://hi-star.etf.bg.ac.rs/ is updated. Website Google Analytics and social networks KPIs are followed regularly. Five journal papers are published in the international journal from the JCR list, and nine conference papers are presented at international conferences. Our paper, presented at the IcETRAN 2023 conference, was selected as the best paper presented in the Section of Telecommunications. Our other paper was selected as the best paper of the young author presented in the same section. Our paper was selected as the best paper in the area of Telecommunications published in a scientific journal in the year 2022/2023. hi-STAR project was presented at the conference Infoteh 2023, East Sarajevo, BiH, at the TELSIKS 2023, Niš, Serbia, at a special session "IoT Applications in Modern and Emerging Technologies" and at the Workshop in Christiаn Doppler Laboratory "Digital Twin assisted AI for sustainable RAN", TUV, Vienna, Austria.
Deliverable 2.1: Hybrid Integrated Satellite and Terrestrial Access Network
This deliverable is a result of the work done in WP2 Subactivity T2.1 – System architecture proposal and state of the art overview. Deliverable D2.1 presents description of network architecture of the overall user access that comprises the user terminal, 5G and satellite RATs and gateway as a merge point. The selected network architecture is explained and its selection is justified. Also, for radio access parts of the architecture proper channel models and simulation environments are elaborated.
Deliverable 2.2: HUT architecture
This deliverable is a result of the work done in WP2 Subactivity T2.1 – System architecture proposal and state of the art overview. Deliverable D2.2 presents hybrid user terminal (HUT) architecture. It relies on the previous deliverable D2.1 and network architecture for hybrid access that was presented in D2.1. Based on that network architecture, HUT system architecture is developed. In this deliverable, overall HUT system architecture is presented and explained. Architecture choices regarding implementation aspects are explained and justified.
Deliverable 2.3: Reliability analysis of 5G/Sat hybrid network
This deliverable is a result of the work done in the context of WP2 Subtask T2.2 (Reliability analysis of integrated terrestrial and satellite links). We briefly presented the outage analysis of the hybrid integrated satellite-terrestrial communication systems. First, we discussed the case where the reliability of the system is increased by using the cooperative relaying, and the analysis takes into account the system geometry and realistic channel parameters. Finally, we present a novel framework for designing decoders, for Low-Density Parity-Check (LDPC) codes, which surpasses the frame error rate performance of Belief-Propagation (BP) decoding. Its key component is the adaptation method, based on the genetic optimization algorithm.
Deliverable 3.1: Unified SDR framework for DVB-S2X and 5G modems
This deliverable is a result of the work done in the context of WP3 Subtask T3.1 – Analysis of existing SDR frameworks for DVB-S2X and 5G. Deliverable D3.1 presents description of physical layer architecture for DVB-S2X and 5G modems. Deliverable D3.2 will present developed FPGA hardware accelerators for both physical layers.
Deliverable 3.2: FPGA-based accelerators for high-performance 5G/Sat transceivers
This deliverable is a result of the work done in the context of WP3 Subtask T3.2 – 5G and DVB-S2X protocol implementation using HW/SW co-design. Deliverable D3.2 presents developed FPGA hardware accelerators for both physical layers.
Deliverable 4.1: Traffic control unit based on deep neural network learning
This deliverable is a result of the work done in the context of WP4 Subtasks T4.1 (Simulation environment creation) and T4.2 (Design of traffic control module placed in HUT). We explain the simulation environment that includes LEO (Low Earth Orbit) satellite-to-ground communication links and terrestrial millimeter-wave links. Then, we propose a user-centric handover execution method that measures instantaneous signal-to-noise ratio and incorporates deep neural networks to predict channel state information and reduce the data loss probability of the transmission.
Deliverable 4.2: Simulation and verification of the ITCU module
This deliverable is a result of the work done in the context of WP4 Subtasks T4.1 (Simulation environment creation) and T4.2 (Design of traffic control module placed in HUT). We briefly review the simulation environment that includes LEO (Low Earth Orbit) satellite-to-ground communication links and a user-centric handover execution method. Following this, a description of the expert system based on various machine learning (ML) models including (neural networks) is given, alongside the dataset containing the simulations for its evaluation. A comparison between various ML models is given, and a review of system performance in terms of two types of adaptive coding and modulation scenarios.
Deliverable 5.1: Functional network core gateway for PoC
This deliverable is a result of the work done in WP5 Subactivity T5.1 – Network core functions and gateway design and development. Deliverable D5.1 presents functional gateway that connects to network cores of terrestrial and satellite RATs to enable ATSSS (Access Traffic Steering, Switching and Splitting) functions to users that have possibility on their devices to connect to these two RATs. The purpose of the gateway is to enable persistent, robust and reliable TCP (Transmission Control Protocol) sessions to users prone to handovers and/or loss of connection over one of the RATs. This deliverable also discusses UDP (User Datagram Protocol) traffic from ATSSS and end user points of view. Discussion regarding possible approaches in enabling ATSSS support is given as well.
Deliverable 5.2: Edge control center for optimal user traffic distribution
This deliverable is a result of the work done in WP5 Subactivity T5.2 – Network core edge functions research and development. Deliverable D5.2 presents architectural and functional survey and analysis of edge computing used for enhancing quality of service and experience (QoS, QoE) for end users. Edge computing has been recognized as important pillar for improving quality of provided services and meeting high demands especially regarding very low latency and bandwidth efficient usage. In this deliverable, we discuss the possibilities to use edge computing to optimally distribute user traffic by notifying users which RAT is more suitable for specific services.
Deliverable 5.3: Use cases and business models
This deliverable D5.3 is a result of the work done in the context of WP5 Subtask T5.3 – Use cases definition and development of business models. The deliverable provides description and detailed technical analysis of selected use cases as well as report on business opportunity analysis and business models development.
Deliverable 7.1: Website, communication channels and project dissemination materials
The hi-STAR project addresses one of the most critical challenges for the next generation wireless networks, which is integration of non-terrestrial networks with terrestrial 5G network. The general objective of the project is to develop flexible framework for integrated terrestrial 5G and Low-Earth-Orbit (LEO) satellite networks, where traffic management is performed with assistance of newly developed artificial intelligence methods. This deliverable is a result of the work done in the context of WP7 Subtask T7.1 – Website, communication channels and project dissemination materials. Deliverable D7.1 presents description of developed project website and project dissemination and communication plan. Deliverables D7.2 and D7.3 will present Mid-term report on dissemination and communication activities (M1-M18) and Final report on dissemination and communication activities, respectively (M19-M36).
Deliverable 7.2: Mid-term report on dissemination and communication activities
This deliverable is a result of the work done in the context of WP7 Subtask T7.1 – Website, communication channels and project dissemination materials and WP7 Subtask T7.2 – Publishing results at journals and conferences. While in the deliverable D7.1 is presented description of developed project website and project dissemination and communication plan, this deliverable D7.2 present Mid-term report on realized dissemination and communication activities in the first period of the project (M1-M18). Consequently, deliverable D7.3 will be Final report on dissemination and communication activities in the second period of project and will be an update of D7.2 deliverable for the period (M19-M36).