Dissemination

Project Meetings and Dissemination Events

Participation in IEEE GLOBECOM 2021

The technical manager Dr. Albena Mihovska participated in the Idustry Panel "5G AND BEYOND - THE PERSPECTIVE OF EUROPEAN RESEARCH PROJECTS"

Motor5G Kick-off Meeting

The Kick-off meeting of the Motor5G project was held on December 9-10, 2019 in Brussels, Belgium

Publications

  • H. Al Kassir, Z. D. Zaharis, P. I. Lazaridis, N. V. Kantartzis, T. V. Yioultsis, and T. D. Xenos ‘’ A Review of the State of the Art and Future Challenges of Deep Learning-Based Beamforming ‘’ Submitted to IEEE ACCESS (under revision)

Abstract: The key objective of this paper is to explore the recent state-of-the-art artificial intelligence (AI) applications on the broad field of beamforming. Hence, a multitude of AI-oriented beamforming studies are thoroughly investigated in order to correctly comprehend and profitably interpret the AI contribution in the beamforming performance. Starting from a brief overview of beamforming, including adaptive beamforming algorithms and direction of arrival (DOA) estimation methods, our analysis probes further into the main machine learning (ML) classes, the basic neural network (NN) topologies, and the most efficient deep learning (DL) schemes. Subsequently, and based on the prior aspects, the paper explores several concepts regarding the optimal use of ML and NNs either as standalone beamforming and DOA estimation techniques or in combination with other implementations, such as ultrasound imaging, massive multiple- input multiple-output structures, and intelligent reflecting surfaces. Finally, particular attention is drawn on the realization of beamforming or DOA estimation setups via DL topologies. The survey closes with various important conclusions along with an interesting discussion on potential future aspects and promising research challenges.


  • H. Al Kassir, Z. D. Zaharis, P. I. Lazaridis, N. V. Kantartzis, T. V. Yioultsis, I. P. Chochliouros, A. Mihovska, and T. D. Xenos ‘’ Antenna Array Beamforming Based on Deep Learning Neural Network Architectures ‘’ Accepted at 3rd URSI AT-AP-RASC, Gran Canaria, 29 May – 3 June 2022.

Abstract: The implementation of antenna array beamforming using several neural network (NN) architectures is compared in this paper. Gated recurrent unit, feed-forward NN, convolutional NN, and long short-term memory architectures have been used for the beamforming process. This comparative study is carried out using various metrics, such as the root mean square error, and the computational time for each NN. In addition, the mean absolute divergences of the antenna array main lobe and nulls directions from their respective desired directions have also been used to assess the performance of each beamformer. The neural networks are trained using the simulation results of a 16-element microstrip patch antenna array. It is demonstrated that deep learning-based beamformers are capable of computing optimum antenna array weights in real time and in environments that change with time.


  • H. Al Kassir, Z. D. Zaharis, P. I. Lazaridis, N. V. Kantartzis, T. V. Yioultsis, I. P. Chochliouros, and T. D. Xenos ‘’Comparative Study of Neural Network Architectures Applied to Antenna Array Beamforming ‘’ Accepted at IEEE International BlackSeaCom Conference, Sofia, Bulgaria, 6–9 June 2022.

Abstract: A comparison of various neural network (NN) architectures is performed in this paper in order to be used as beamformers applied to a linear antenna array composed of 16 microstrip elements. Two recurrent NNs using respectively gated recurrent units and long short-term memory, a convolutional NN, and a feed-forward NN are used here as adaptive beamformers. Three cases are investigated, each one with a different number of incoming signals received by the antenna array, and the performance of each NN structure is evaluated using various metrics. The simulation results demonstrate the effectiveness of the deep learning-based beamformers in real-time calculation of the optimal antenna array weights, while considering ever-changing environments.


  • Ahmed M. Nor, Simona Halunga, Octavian Fratu “Optimal Placement of Two IRSs in Beyond 5G Network” Accepted at 6th EAI International Conference on Future Access Enablers of Ubiquitous and Intelligent Infrastructures (EAI FABULOUS 2022) - May 2022, Zagreb, Croatia.

Abstract: Intelligent reflecting surfaces (IRSs) become a major player in beyond 5G networks because it can assist millimeter wave and terahertz bands to over-come their propagation and blockage issues. IRSs provide network with alterna-tive line of sight paths and extend the network coverage. However, deploying a single IRS seems not enough specially in crowded large indoor area with block-ing scenarios. In this paper, we discuss the scenario of the network with two im-plemented IRSs and compare it with the case of single IRS hence proving the superiority of first case to extend the coverage and reduce the probability of blockage occurrence in the network. Then, we propose the optimal placement of two IRSs in the environment to enhance the overall performance of the network. The proposed method is based on maximizing the average received power overall possible user equipment (UEs) positions within the study area. This method can guarantee larger received power for almost all UE positions. Finally, we study the performance of the network with different blockage probability of occurrence cases not only to the direct link between access point and UE but also to the link between IRS and UE.


  • Ahmed M. Nor, Octavian Fratu, Simona Halunga, Ayham Alyosef, Zaharias D. Zaharis, Stamatia Rizou, Pavlos I. Lazaridis “Demand based Proportional Fairness Scheduling for 5G eMBB Services” Accepted at IEEE BlackSeaCom conference, Sofia, Bulgaria, 6. June – 9 June 2022.

Abstract: Millimeter wave (mmWave) based 5G networks have nowadays attracted researcher from both industry and academia. Because, mmWave can enable dominant 5G services such as ultra-high definition video streaming and virtual reality, by providing them with their requirements, e.g., high data rate and low latency time. But, mmWave channel dynamically changes, which effects on overall performance of 5G network. Hence, efficient radio resource management schemes are needed in medium access control (MAC) layer to wisely distribute mmWave band resources among user equipment (UEs). In this paper, we design a high performance MAC scheduling scheme by improving standard proportional fairness (SPF) scheduling algorithm. In which, UEs demands are considered besides their channel condition. The proposed demand based proportional fairness (D-PF) scheme firstly gives higher priority to UEs with lower data rates requirements and better channel condition, hence higher UEs satisfaction level can be obtained faster. Then, it prioritizes them only based on their channel qualities if their required rates are achieved. Through simulation, we prove that the proposed scheme outperforms the conventional round robin (RR) and SPF algorithms in terms of system throughput, where an increase of more than 300 Mbps can be achieved, with maintaining same fairness between UEs. Meanwhile, D-PF scheme provides better throughput and satisfaction for UEs. For example, it can satisfy 95% of network UEs by providing them with their required rates, while only 61% and 62% of UEs can be satisfied if RR or SPF is used, respectively.

  • Ayham Alyosef, Stamatia Rizou, Zaharias D. Zaharis, Pavlos I. Lazaridis, Ahmed M. Nor, Octavian Fratu, Simona Halunga, Traianos V. Yioultsis, Nikolaos V. Kantartzis “A Survey on the Effects of Human Blockage on the Performance of mmWave Communication Systems” Accepted at IEEE BlackSeaCom conference, Sofia, Bulgaria, 6. June – 9 June 2022.

Abstract: Human blockage is one of the key challenges that limit the ability of mmWave communications to provide ultra-high data rate and ultra-low latency links, thus severely reducing the quality-of-service (QoS) experienced by the users. In this paper, we present the most common human body blockage models, which are used in blockage analysis to predict the level of attenuation caused by the human body to mmWave signals. Moreover, the main parameters of human blockage which affect the received signal are discussed, while the effect of blockage on the received signal and network coverage is analyzed. Finally, we provide insights to potential solutions that overcome human blockage, in order to further improve the overall performance of mmWave communications.


  • Ahmed M. Nor, “Joint Proportional Fairness Scheduling Using Iterative Search for mmWave Concurrent Transmission” Accepted at IEEE BlackSeaCom conference, Sofia, Bulgaria, 6. June – 9 June 2022.

Abstract: Millimeter wave (mmWave) will play a significant role as a 5G candidate in facing the growing demand of enormous data rate in the near future. The conventional mmWave standard, IEEE 802.11ad, considers establishing only one mmWave link in wireless local area network (WLAN) to provide multi Gbps data rate. But, mmWave has a tenuous channel which hinders it from providing such rate. Hence, it's necessary to establish multiple mmWave links simultaneously by deploying a multiple number of mmWave access points (APs) in 5G networks. Unfortunately, applying conventional standard without any modifications for mmWave concurrent transmission impedes mmWave APs from selecting optimum mmWave concurrent links. Because IEEE 802.11ad standard associates the user equipment (UEs) to mmWave APs using the link that has the maximum received power without considering mutual interference between simultaneous links. In this paper, a joint proportional fairness scheduling (JPFS) optimization problem for establishing optimum mmWave concurrent transmission links is formulated. And, to find a solution to this non-polynomial (NP) time problem, we use exhaustive search (ES) scheme. Numerical simulation proves the effectiveness of using the ES scheme to improve the system performance.


  • A. U. Haq, M. Bilal, A. Mihovska, “Resource Allocation Ensuring Physical Layer Security in Cooperative Non-Orthogonal Multiple Access in 6G Networks”, in IEEE Blackseacom Conference, Bulgaria, June 2022

Abstract: Non-Orthogonal Multiple Access (NOMA) is a potential candidate for 6th Generation (6G) cellular communication. It outperforms the conventional Orthogonal Multiple Access (OMA) schemes in terms of connectivity, spectral efficiency and latency. In spite of these advantages, NOMA faces information security challenges. In this paper, two phases of NOMA are considered which are the direct transmission phase and the cooperative phase. In the direct transmission phase, all users receive the superimposed signal while in the cooperative phase the user with a higher channel condition helps the user with the weaker channel condition. In the proposed work, the secrecy rate of users with weaker channel conditions is considered for maximisation under total power and Quality-of-Service (QoS) constraints. Our results show an improvement in the secrecy rate.

  • Sefati, S. S., & Tabrizi, S. G. (2021). Detecting Sybil Attack in Vehicular Ad-hoc Networks (Vanets) by Using Fitness Function, Signal Strength Index and Throughput. Wireless Personal Communications, 1-21.

Abstract: Ad-hoc networks are vehicular networks whose functions are widely expanding. High dynamicity and the presence of wireless communications are two major challenging features of these networks. Due to, security and protection of these networks have remained as big unresolved challenges. Attack detection in Vanets may have a significant impact on the efficiency of these networks. Timely detection of attacks helps prevent road casualties and traffic control. Initial identification is done by the neighbouring nodes. When each node receives a message from neighbouring nodes, it compares the ID of each node with those of other nodes’ messages. If the messages are the same but sent from different nodes, neighbouring nodes send a sample of the data to the RSU. If RSU doubts to an ID, it establishes a table of parameters, such a delay, packet drop and throughput. The total sum of fitness functions should be equal to one. If the fitness function of these parameters is beyond the determined limit, nodes will continue to work. In this paper detected Sybil attacks used by the NS3 simulator. The results of the proposed method were acceptable in comparison with PDF (probability density function), FCVS (Fuzzy-based collaborative verification system) and Heartbeat scheme. Delay, energy and strength were the parameters, which had higher responsiveness.


  • Sefati, S. S., Halunga, S., & Farkhady, R. Z. (2022). Cluster selection for load balancing in flying ad hoc networks using an optimal low-energy adaptive clustering hierarchy based on optimization approach. Aircraft Engineering and Aerospace Technology, (ahead-of-print).


Abstract: Purpose– Flying ad hoc networks (FANETs) have a major effect in various areas such as civil projects and smart cities. The facilities of installation and low cost of unmanned aerial vehicles (UAVs) have created a new challenge for researchers. Cluster head (CH) selection and load balancing between the CH are the most critical issues in the FANETs. For CH selection and load balancing in FANETs, this study used efficient clustering to address both problems and overcome these challenges. This paper aims to propose a novel CH selection and load balancing scheme to solve the low energy consumption and low latency in the FANET system.

Design/methodology/approach – This paper tried to select the CH and load balancing with the help of low-energy adaptive clustering hierarchy (LEACH) algorithm and bat algorithm (BA). Load balancing and CH selection are NP-hard problems, so the metaheuristic algorithms can be the best answer for these issues. In the LEACH algorithm, UAVs randomly generate numerical numbers, and these numbers are sorted according to those values. To use the load balancing, the threshold of CH has to be considered; if the threshold is less than 0.7, the BA starts working and begins to find new CH according to the emitted pulses.

Findings – The proposed method compares with three algorithms, called bio-inspired clustering scheme FANETs, Grey wolf optimization and ant colony optimization in the NS3 simulator. The proposed algorithm has a good efficiency with respect to the network lifetime, energy consumption and cluster building time.

Originality/value – This study aims to extend the UAV group control concepts to include CH selection and load balancing to improve UAV energy consumption and low latency.


  • Sefati, S. SHalunga, Mobile sink assisted data gathering for URLLC in IoT using a fuzzy logic system. “IEEE Black sea conferences 2022”


Abstract: The Internet of Things (IoT) is a new technology that employs a variety of sensors and wireless communication protocols. People are leveraging IoT to facilitate their actions by using innovative and intelligent equipment. An increasing number of sensors can decrease the Quality of Service (QoS), so implementing a suitable algorithm in an IoT network may enhance this parameter. One of the important applications in IoT with strict QoS requirements is the Ultra-Reliability and Low Latency Communication (URLLC). Several types of URLLC services can be provided in the IoT based on the user’s requirements. Also, a proper Cluster Head (CH) selection plays a vital role in receiving and transmitting data and reduces the energy consumption of the devices. In today's IoT devices, the mobility of the sink nodes is one of the most important features to be considered in real-world applications. The selection of the CH is an NP-Hard problem; thus, a fuzzy logic protocol is used to select the appropriate CH and intermediate nodes for routing, and the results are presented in this paper. This method is developed in order to minimise the delay, and energy consumption as well as enhancing the reliability. The method has been tested via MATLAB Simulation and the results are presented in the final section of the present paper.


  • Sefati, S. S Halunga, Data Forwarding to FoG With Guaranteed Fault Tolerance in Internet of Things (IoT). “COMM2022”

Abstract: The Internet of Things (IoT) is a technology that employs a variety of sensors and wireless communication protocols. People are leveraging IoT to make their lives easier by using innovative and intelligent equipment. One of them is home automation, which works in conjunction with the actuators and sensors linked to the network. The massive increase in the number of sensors, embedded systems, personal devices, etc., will increase the load and, thus, the faults. In such complicated and dense networks, the capacity to offer reliable services in any circumstance is required, and the IoT sensors should not be interrupted. Fault-tolerance methods are essential to maintain low latency and high reliability in IoT systems. A dynamic and sophisticated algorithm is needed to ensure the services in real-world applications. In this proposed method, firstly, the Fog retrieves and stores the Availability, Reliability, and Throughput information (ART) from each cluster. Then, the second round compares the parameters available for the current situation because it is possible to fail in each round. If the ART is equal with the first round, the situation is continuous; otherwise, Fog requests from the Cluster Head (CH) to find the faulty sensor. By removing the faulty sensor, the adjacent nodes must re-route. The proposed method has been simulated in a NS3 environment and achieved high reliability and throughput compared to other algorithms.


  • Pablo H. Zapata-Cano, E. Vassos, Z. D. Zaharis, P. I. Lazaridis, N. V. Kantartzis, T. V. Yioultsis, N. V. Kantartzis, and A. Feresidis ‘’Optimization-driven design of a 90◦ metasurface phase shifter at 60 GHz ‘’ at IEEE International BlackSeaCom Conference, Sofia, Bulgaria, 6–9 June 2022.

Abstract: The exigent demands imposed by future generations of high-speed cellular and satellite communications claim for low-loss and wideband phase shifters at mm-Wave frequencies. Pixelated metasurfaces provide large design versatility and constitute an attractive solution for wave manipulation. However, their design often implies the simultaneous tuning of a large number of geometrical parameters. In this article, multiobjective optimization is used together with full-wave simulation to design a low-loss 90◦ phase shifter operating on the 57-63 GHz frequency band. Among the set of optimal individuals provided by the algorithm, a final solution has been selected according to the electromagnetic response of the device, achieving less than 0.06 dB of reflection loss and a constant phase shift with an absolute error less than 2◦ over the whole frequency band.

  • Pablo H. Zapata-Cano, Z. D. Zaharis, N. V. Kantartzis, T. V. Yioultsis, and C. Antonopoulos ‘’On the Utilization of Graphene Derivatives for Microwave Applications ‘’ at 20th International Symposium on Applied Electromagnetics and Mechanics (ISEM) 2022.

Abstract: Thanks to its remarkable properties, graphene has provoked the interest of research community in the last years. Moreover, the simple top-down synthesis of chemically modified graphene materials, such as graphene oxide (GO) or reduced graphene oxide (rGO), opens new possibilities for an easier and more feasible fabrication. However, relatively few applications have been proposed and realistically assessed for microwave devices. Here we provide a brief overview of the main enablers and challenges of the utilization of graphene derivatives for microwave applications.


  • G. Kougioumtzidis, V. Poulkov, Z. Zaharis and P. Lazaridis, “Machine Learning for QoE Management in Future Wireless Networks”, 2021 XXXIVth General Assembly and Scientific Symposium of the International Union of Radio Science (URSI GASS 2021), Rome, Italy, 28 Aug. -4 Sep. 2021, pp. 1-4, doi: 10.23919/URSIGASS51995.2021.9560226.

Abstract: The growth in volume and heterogeneity of accessible services in future wireless networks (FWNs), imposes pressure to communication service providers (CSPs) to expand their capacity for network performance monitoring and evaluation, in particular in terms of the performance as it is perceived by end-users. The quality of experience (QoE)-aware design model allows to understand and analyze the operation of networks and services from the end-user’s perspective. In addition, network measurements based on QoE constitute a key source of knowledge for the overall functionality and management of the network. In this respect, the implementation of artificial intelligence (AI) and machine learning (ML) in QoE management, increases the accuracy of modeling procedures, improves the monitoring process efficiency, and develops innovative optimization and control methodologies.

  • G. Kougioumtzidis, V. Poulkov, Z. Zaharis and P. Lazaridis, “A Survey on Multimedia Services QoE Assessment and Machine Learning Based Prediction”, IEEE Access, vol. 10, pp. 19507-19538, 2022, doi: 10.1109/ACCESS.2022.3149592.

Abstract: The groundbreaking evolution in mobile and wireless communication networks design in recent years, in combination with the advancement of mobile terminal equipment capabilities, has led in an exponential growth of mobile internet technologies, and arose an ever-growing demand for innovative multimedia services. The highly demanding in terms of network resources over-the-top media services, as well as the emergence of new and complex mobile multimedia services such as video gaming, ultra-high-definition video, and extended reality, requires the enhancement of end-users’ perceived quality of experience (QoE). QoE has garnered much research interest in recent years, and has emerged as a key component in the evaluation of network services and operations. As a result, a QoE-aware network planning approach is getting increasingly favored, and novel design challenges, such as how to quantify and measure QoE, have arisen. In this regard, a paradigm shift in network implementations is being envisioned, in which the focus will be on machine learning (ML) methodologies for developing QoE prediction models, directly related to end-user’s personalized experience. In this survey, an analysis on application-oriented, ML-based QoE prediction models for the goal of QoE management for multimedia services is presented. In addition, an examination of the state-of-the-art ML-based QoE predictive models and some of the innovative techniques and challenges related to multimedia services quality assessment with focus on extended reality and video gaming applications are outlined.

  • G. Kougioumtzidis, V. Poulkov, Z. Zaharis and P. Lazaridis, “QoE Assessment Aspects for Virtual Reality and Holographic Telepresence Applications”, Accepted at 6th EAI International Conference on Future Access Enablers of Ubiquitous and Intelligent Infrastructures (EAI Fabulous 2022), Zagreb, Croatia, May 4-6, 2022.

Abstract: The cutting-edge evolution of mobile communication systems and Internet technologies in nowadays transitional period from the information age to the experience age has brought attention to the evolving virtual reality (VR) and augmented reality (AR) applications and moves towards the development of holographic telepresence systems. Since these applications are devoted in creating immersive and interactive experiences, the quality of experience (QoE) as it is perceived by the end-users will become fundamental constituent in their performance evaluation process. In this paper, the significance of QoE in the development and implementation of the emerging technologies of VR and holographic telepresence systems is analyzed. Moreover, the QoE influencing factors for VR applications and the distinction among this evolv-ing technology and the conventional 2D video content are outlined. Furthermore, a classification of the QoE assessment methods, together with an analysis of the more significant metrics with regard to VR applications is presented.

  • G. Kougioumtzidis, V. Poulkov, Z. Zaharis, and P. Lazaridis, "Intelligent and QoE-aware Open Radio Access Networks", Accepted at 2022 3rd URSI Atlantic Radio Science Meeting (AT-AP-RASC 2022), Gran Canaria, Spain, May 29- Jun. 3, 2022.

Abstract: The open radio access network (O-RAN) concept refers to the architectural design of the next generation RAN that is built on the principles of openness and intelligence. This paper presents an overview of the concept of O-RAN, by analyzing its architecture and examining its main building blocks. Moreover, it highlights the significance of the quality of experience (QoE) for the envisaged future wireless networks, and analyzes the importance of integrating QoE-awareness in O-RAN's design. Furthermore, it provides an analysis of the methodology of embedding artificial intelligence models in O-RAN's architecture with the form of xApps.

  • C. Milias, R. Andersen, P. Lazaridis, Z. Zaharis, B. Muhammad, J. Kristensen, A. Mihovska and D. Hermansen, “Metamaterial-inspired Antennas: A Review of the State of the Art and Future Design Challenges”, IEEE Access, vol. 9, pp. 89846-89865, 2021, doi: 10.1109/ACCESS.2021.3091479

Abstract: Metamaterials are artificial structures with the ability of exhibiting unusual and exotic electromagnetic properties such as the realisation of negative permittivity and permeability. Due to their unique characteristics, metamaterials have drawn broad interest and are considered to be a promising solution for improving the performance and overcoming the limitations of microwave components and especially antennas. This paper presents a detailed review of the most recent advancements associated with the design of metamaterial-based antennas. A brief introduction to the theory of metamaterials is provided in order to gain an insight into their working principle. Furthermore, the current state-of-the-art regarding antenna miniaturisation, gain and isolation enhancement with metamaterials is investigated. Emphasis is primarily placed on practical metamaterial antenna applications that outperform conventional methods and are anticipated to play an active role in future wireless communications. The paper also presents and discusses various design challenges that demand further research and development efforts.

  • C. Milias, R. Andersen, P. Lazaridis, Z. Zaharis, B. Muhammad, J. Kristensen, A. Mihovska and D. Hermansen, “End-fire antenna array with metamaterial decoupling structures for UAV-borne radar”, Accepted at 2022 3rd URSI Atlantic Radio Science Meeting (AT-AP-RASC 2022), Gran Canaria, Spain, May 29- Jun. 3, 2022.

Abstract: This paper presents the design and implementation of an X-band, 2D, end-fire antenna array that is suitable for UAV-borne radar applications. The array is synthesised by eight printed Yagi-Uda individual antennas arranged in a 4 x 2 configuration and has a 17 dBi gain. Each antenna is fed by a microstrip balun and consists of one driven dipole, one reflector and six directors. Since mutual coupling is detrimental to the performance of phased arrays, metamaterial decoupling structures are utilised to enhance the isolation between them. More specifically, mu-negative meanders are inserted between the elements in the E-plane, while a metasurface of ring resonators isolates the elements in the H-plane. As a result, the inter-element coupling is lower than -30 dB. The proposed platform can be easily mounted on unmanned aerial vehicles (UAVs) and is an ideal candidate for UAV-borne radars. Simulations along with measurements verify the effectiveness of our design.

  • C. Milias, R. Andersen, P. Lazaridis, Z. Zaharis, B. Muhammad, J. Kristensen, A. Mihovska and D. Hermansen, “Miniaturized Multi-Band Metamaterial Antennas with Dual-Band Isolation Enhancement”, IEEE Access, Accepted.

Abstract: We present electrically small, multi-band, metamaterial-inspired antennas with adequate radiation characteristics and isolation enhancement. The antenna element consists of a complementary split-ring resonator (CSRR) embedded in a small monopole that has a size of λ/8 x λ/10 at the lowest frequency band, while rectangular patches are placed underneath it to further improve the performance. The antenna operates at the 2.4-2.5/2.9-4.8/5.1-6.5 GHz frequency bands. Moreover, we propose a systematic, metamaterial-based approach in order to improve the isolation between two of these small, closely spaced antenna elements at the lowest and highest frequency bands. The proposed techniques reduce the coupling by up to 29 dB without increasing the size of the structure. In particular, the isolation enhancement at the highest frequency band of interest is remarkably wideband. The cable effect, which is a common concern during the measurements of small antennas, is examined as well. The proposed antennas are not only small but also densely packed and can be easily integrated with modern, compact communication devices with advanced functionality. Simulations along with experimental results validate the effectiveness of our design.

  • I. Mallioras, Z. D. Zaharis, P. I. Lazaridis and S. Pantelopoulos, "A Novel Realistic Approach of Adaptive Beamforming based on Deep Neural Networks," in IEEE Transactions on Antennas and Propagation, doi: 10.1109/TAP.2022.3168708.

Abstract: A new deep neural network (NN) approach applied to antenna array adaptive beamforming is presented in this paper. A recurrent NN (RNN) based on the gated recurrent unit (GRU) architecture is used as a beamformer in order to produce proper complex weights for the feeding of the antenna array. The proposed RNN utilizes four hidden GRU layers and one extra layer for linear transformation. The produced weights are subsequently compared with respective weights derived by a null steering beamforming (NSB) technique in order to measure the accuracy of the RNN. The RNN training is performed by using a large data set derived from an NSB technique applied to a realistic microstrip linear antenna array, in order to take into account real-world effects, like the non-isotropic radiation pattern of an array element and the mutual coupling between the array elements. The RNN performance is examined by using the root mean square error metric, whereas its beamforming performance is evaluated by estimating the mean value and the standard deviation of the divergences of the main lobe and nulls directions from their respective desired directions. A comparison between various NN structures and an overall study of the proposed RNN-based beamformer are also presented.

  • I. Mallioras, Z. Zaharis, P. Lazaridis, I. Chochliouros, K. Mistry, T. Loh, “A Novel Approach based on Recurrent Neural Networks Applied to Adaptive Beamforming” at 23rd Conference on the Computation of Electromagnetic Fields COMPUMAG 2021, 20 Jan. 2022

Abstract: Embedding artificial intelligence into current communication systems will substantially assist the development of ecosystems beyond 5G. In this study, a new neural network (NN) approach applied to antenna array adaptive beamforming is presented. A Recurrent NN (RNN) based on the Gated Recurrent Unit (GRU) architecture is used as a beamformer in order to receive the angles of arrival (AoA) of incoming signals and produce the complex weights for the feeding of the elements of the antenna array. These weights are subsequently compared with the respective weights derived by a null steering beamforming (NSB) algorithm to measure the accuracy of the RNN. The proposed RNN utilizes four hidden GRU layers and one extra layer for linear transformation. The RNN training is performed by using a large data set derived from a realistic antenna array using NSB as the beamforming technique. The RNN performance is tested using the root mean square error (RMSE) metric, whereas its beamforming performance is evaluated by estimating the mean deviation of the main lobe and null directions from their respective desired directions. A comparison between various NN structures and an overall study of the proposed RNN-based beamformer are also presented.

  • I. Mallioras, Z. D. Zaharis, P. I. Lazaridis, N. V. Kantartzis, T. V. Yioultsis, Bo Liu, Stavros Kalafatis, “A Novel Utilization of NARX for Antenna Array Adaptive Beamforming” at 3rd URSI Atlantic Radio Science Meeting (AT-AP-RASC 2022), Gran Canaria, Spain, May 29- Jun. 3, 2022.

Abstract: In this paper, we investigate the use of a nonlinear autoregressive network with exogenous inputs (NARX) for adaptive beamforming on smart antennas. As a beamformer, NARX receives the angles of arrival of incoming signals to extract the complex feeding weights that produce the appropriate antenna radiation pattern. In order to demonstrate the potential of such an implementation, we test our model on a realistic linear antenna array composed of 16 microstrip elements. We use the null steering beamforming technique to produce the datasets needed for training and testing of our model and then we evaluate the accuracy of the radiation patterns produced by this model. To further demonstrate the efficiency of the NARX implementation, we also make a comparison with a feed-forward neural network that has the same architecture with that of NARX.

  • I. Mallioras, Z. D. Zaharis, P. I. Lazaridis, V. Poulkov, N. V. Kantartzis, T. V. Yioultsis, “An Adaptive Beamforming Approach Applied to Planar Antenna Arrays Using Neural Networks” at IEEE International BlackSeaCom Conference, Sofia, Bulgaria, 6–9 June 2022.

Abstract: Future wireless networks depend on the improvement of current smart antenna operations so that they maintain high accuracy levels at low response times. Utilizing machine learning techniques, it is possible to replace the currently used algorithms with a much faster yet reliable alternative. In this study, we focus on adaptive beamforming applied to a planar antenna array using the null steering beamforming algorithm (NSB). We test different types of deep neural networks (DNNs) as potential alternative beamformers, by comparing their accuracy to that of the NSB algorithm. The application concerns an 8×8 planar antenna array composed of isotropic elements. The DNNs tested here are the traditional feedforward neural networks, the non-linear autoregressive networks with exogenous inputs, and recurrent neural networks using either gated recurrent units or long short-term memory units. In addition, we investigate each DNN type to make sure we are utilizing the best version of each neural network architecture.


  • I. Mallioras, Z. Zaharis, P. Lazaridis, N. V. Kantartzis, T. V. Yioultsis, I. Chochliouros, “Comparative study of a deterministic adaptive beamforming technique with Neural Network implementations” at 18th International Conference on Artificial Intelligence Applications and Innovations AIAI 2022, Crete, Greece, 17-20 June, 2022

Abstract: Future wireless networks depend on the development of new mechanisms that can increase the efficiency of the network. Antenna array adaptive beamforming (ABF) is an antenna operation that can be significantly improved with the use of machine learning. In this paper, a deterministic beamforming technique is compared with two different types of neural networks (NNs). These are the non-linear autoregressive network with exogenous inputs (NARX) and the recurrent NN (RNN) with long short-term memory (LSTM) units. To train the NNs, we produce a dataset using the minimum variance distortionless algorithm (MVDR) applied to a realistic antenna array. Using grid search, we find the best architecture for both NNs. Then, we train the final models and evaluate them by comparing their accuracy to that of the MVDR algorithm. We demonstrate how the use of NNs is preferable to that of deterministic algorithms as they appear to maintain high accuracy while having a much lower response time than that of deterministic algorithms. The RNN with LSTM units is the most promising out of the two NN models as it achieves higher accuracy with a slightly shorter training time