| Short name | Deliverable | Abstract |
|---|---|---|
| SORUS-DRONE-A1.1-E1 | First version of the architecture | The main objective of the three 6G-SORUS projects to study the integration of UAVs with RIS and vRAN. This integration can provide several benefits, including increased coverage and capacity, improved spectral and energy efficiency, and enhanced security and privacy. UAVs can be used as flying base stations to provide wireless coverage to areas where it is difficult or impossible to deploy traditional base stations. By using RIS, the UAVs can enhance the signal quality and coverage, while vRAN can provide dynamic resource allocation and efficient management of the wireless resources. This document describes the architectural elements for the integration of UAVs in the operation of the network. |
| SORUS-DRONE-A1.2-E1 | UAV Use Cases | The SORUS-DRONES project focuses on the challenges posed by the integration of unmanned aerial vehicles (UAV, UAS, drones) with virtualized radio access networks (vRAN) and reconfigurable intelligent surfaces (RIS) in the deployment and operation of a B5G network. The design and development of the UAV solution will be carried out taking into account the needs of the use cases that will allow the validation of the developments achieved in the other subprojects (vRAN and RIS), as well as the state of the art of technology related to drones, communications and AI-Edge Computing. The proposed solution will address the functionalities of the different use cases with a sufficient degree of flexibility and adaptability that will allow its operational implementation for demonstrations. |
| SORUS-DRONE-A2.1-E1 | UAV Profiling - Initial version | The SORUS-DRONES project focuses on the challenges posed by the integration of unmanned aerial vehicles (UAVs, UAS, drones) with virtualized radio access networks (vRAN) and reconfigurable intelligent surfaces (RIS) in the deployment and operation
of a B5G network. The process of analyzing different solutions and defining prototypes for validation testing is carried out in the PROFILING phase. This document presents the conceptual and functional development of this PROFILING phase, as well as the preliminary results of the initial tests aimed at creating the tools required to achieve the goals of the project. |
| SORUS-DRONE-A2.1-E2 | UAV Profiling - Intermediate Version | The SORUS-DRONES project focuses on the challenges posed by the integration of unmanned aerial vehicles (UAVs, UAS, drones) with virtualized radio access networks (vRAN) and reconfigurable intelligent surfaces (RIS) in the deployment and operation of a B5G network. The process of analyzing different solutions and defining prototypes for validation testing is carried out in the PROFILING phase. This document presents an iteration of the conceptual and functional development of this PROFILING phase, with results of an intermediate stage for the tests aimed at creating the tools required to achieve the goals of the project. |
| SORUS-DRONE-A2.2-E1 (RE10) | Network Optimization I: algorithms for the or- chestration of UAV 6G Networks | The main objective of 6G-SORUS is to study the integration of UAVs with RIS and vRAN. This document describes a set of algorithms and techniques to be used when orchestrating a B5G scenario with drones. The document provides a review of the state of the art and an initial subset of techniques to apply. These are divided in three groups: general algorithms to be used by UAVs (path planning, localization, etc.), orchestration algorithms (i.e., algorithms to control the overall operation of the service), and techniques and algorithms to provide communications using drones. |
| SORUS-DRONE-A3.1-E1 (E12) | Initial Performance Evaluation | In this deliverable, we conduct an in-depth analysis of strategies aimed at enhancing Beyond 5G (B5G) scenarios through the integration of Unmanned Aerial Vehicles (UAVs) and Reconfigurable Intelligent Surfaces (RIS). Our investigation focuses on critical Key Performance Indicators (KPIs) such as coverage probability, end-to-end delay, or power consumption. We systematically explore the dynamic interplay between these metrics under various influential factors shaping the B5G landscape, offering foundational insights into optimizing the deployment of UAVs and RIS. |
| SORUS-DRONE-A3.2-E1 | UAV Design | The SORUS-DRONES project focuses on the challenges posed by the integration of unmanned aerial vehicles (UAVs, UAS, drones) with virtualized radio access networks (vRAN) and reconfigurable intelligent surfaces (RIS) in the deployment and operation of a B5G network. The design and development of the UAV solution will be carried out in accordance with the needs of the use cases that will validate the developments achieved in the other subprojects (vRAN and RIS), as well as the state-of-the-art technology related to drones, communications, and AI-Edge Computing. This document describes the design and elements of the First Prototype to be built for the project. |
| SORUS-DRONE-A3.2-E2 | UAV Initial Prototype | The SORUS-DRONES project focuses on the challenges posed by the integration of unmanned aerial vehicles (UAVs, UAS, drones) with virtualized radio access networks (vRAN) and reconfigurable intelligent surfaces (RIS) in the deployment and operation of a B5G network. The design and development of the UAV solution will be carried out in accordance with the needs of the use cases that will validate the developments achieved in the other subprojects (vRAN and RIS), as well as the state-of-the-art technology related to drones, communications, and AI-Edge Computing. This document describes the specifications and preliminary tests of the First Prototype built for the project. |
| SORUS-RAN-A1.1-E1 | First version of the architecture | The main objective of the three 6G-SORUS projects to study the integration of UAVs with RIS and vRAN. This integration can provide several benefits, including increased coverage and capacity, improved spectral and energy efficiency, and enhanced security and privacy. UAVs can be used as flying base stations to provide wireless coverage to areas where it is difficult or impossible to deploy traditional base stations. By using RIS, the UAVs can enhance the signal quality and coverage, while vRAN can provide dynamic resource allocation and efficient management of the wireless resources. This document describes the architectural elements for the integration of vRAN in the operation of the network. |
| SORUS-RAN-A1.2-E1 | vRAN Use Cases | This document corresponds to deliverable A1.2 - VRAN Use Cases - which is in the initial phase of the tasks described in the technical specifications. It describes three relevant vRAN use cases and studies the parameters of interest to optimize, the metrics to consider, and the different optimization objectives. Following the introduction, the primary goal is to identify scenarios where vRAN virtualization is an advantage over existing deployments. The deployment of multiple virtualized O-DUs on shared computing nodes is studied, as well as the advantages, disadvantages, and challenges of deploying different virtualized O-DUs on the same platform. The technical challenges of deploying multiple O-DUs at the computational level and shared hardware acceleration, as well as associated AI services at the Edge, are also described. The use of both Non-Real Time and Near-Real Time controllers is analyzed to improve resource allocation based on demand. |
| SORUS-RAN-A2.1-E1 | Initial vRAN Profiling | This document corresponds to deliverable A2.1 - vRAN Profiling - which is in the initial phase of the tasks described in the technical specifications. This document defines the elements used in the implementation of two testbeds to conduct various experiments aimed at obtaining energy consumption data. These data include both CPU and device consumption, allowing for separate comparison of energy consumption under the use cases outlined in the SORUS-RAN-A1.2 document. This has made it possible to appreciate, in terms of energy consumption reduction, that the hardware platform used plays a fundamental role in the total consumption, making it essential to select the appropriate platform based on the needs. At the same time, finding the right balance of MCS, airtime, and image resolution (if applicable) in each situation is essential to reduce energy consumption without compromising the end-user experience. |
| SORUS-RAN A2.1-E2 | Final vRAN Profiling | This document summarizes the existing features in vRAN environments and solutions through the proposed O-RAN architecture, which includes both non-real-time and near-real-time intelligent radio control. The problem of noisy neighbors is mainly addressed in this type of deployments with resource sharing. An experimental platform based on Docker is presented with which different experiments have been carried out to find the main reason for the exponential performance degradation with the increase of vBS. It has been concluded that secure computing weakly affects this increase (7%) and context switches account for 43% of the total overhead. However, the main problem lies in cache losses, which increase by 500% with five vBS. |
| SORUS-RAN-A2.2-E1 | First version of joint optimization algorithms | This deliverable deals with algorithms created for the optimisation of IoT networks to provide low-cost, high-bandwidth connectivity. For this purpose, an algorithm based on reconfigurable intelligent surfaces (RISs), called RISMA, has been developed, which jointly optimises the RIS parameters and the beamforming strategy at the transmitter from a theoretical perspective. In turn, an algorithm developed from the previous one, called Lo-RISMA, is presented, which aims to provide a solution to the cases where low resolution RISs are used through the study of the standarized mean square error (SMSE) to achieve an optimisation of the aggregate transmission rate of the system in the downlink. |
| SORUS-RAN-A2.2-E2 | Final version of joint optimization algorithms | This document corresponds to the deliverable SORUS-RAN-A2.2-E2, which is in the final phase of the tasks described in the technical specifications. The objective is to extend the previous work with a multi-user use case, as well as to numerically evaluate both the single-user and multi-user use cases using reconfigurable intelligent surfaces (RISs). The optimization aimed to minimize the sum of mean squared errors across all user nodes. To achieve this, the RIS parameters and the base station precoding are iteratively optimized separately. Additionally, two algorithms have been proposed: RISMA, as a low computational complexity solution to find optimal solutions in real time; and Lo-RISMA, designed for low-resolution metasurfaces, with high computational efficiency, ideal for devices with limited resources. The results show a 40% improvement compared to an MMSE precoder and a gain between 20% and 120% compared to a ZF precoder, depending on the network radius. |
| SORUS-RAN-A2.3-E3 | Final version of the prototype to validate vRAN orchestration | This document corresponds to the deliverable SORUS-RAN-A2.3-E3. The work reported in this deliverable addresses the issue of energy consumption in virtualized radio access networks (RAN). In particular, the use of hardware accelerators (HA) leads to a significant increase in energy consumption, which raises costs and has serious environmental consequences. To solve this problem, we collected data from our experimental platform and compared the performance and energy consumption of an HA (NVIDIA V100 GPU) versus a CPU (Intel Xeon Gold 6240R, 16 cores) for low-power software processing. Based on the obtained data, we designed a strategy to offload traffic to the HAs opportunistically in order to save energy while maintaining reliability. However, this offloading strategy must be configured in near real-time for each base station (BS) sharing common computing resources. This poses a collaborative multi-agent problem where the number of involved agents (BS) can be arbitrarily large and change over time. Therefore, we propose an algorithm called ECORAN, which efficiently solves a problem formulated as a multi-agent contextual bandit. Our solution applies concepts from mean-field theory to be fully scalable. Using a real platform and traces from a production mobile network, we demonstrate that ECORAN can provide up to 40% energy savings compared to the approach currently used by the industry. |
| SORUS-RAN-A3.1-E1 | First version of the prototype to validate vRAN orchestration | Este documento conforma el entregable SORUS-RAN-A3.2-E1, en el que se plantean diferentes algoritmos para dar solución a los tres casos de uso planteados en el entregable SORUS-RAN-A1.2. La finalidad de los mismos es la de explorar distintos escenarios para optimizar el consumo energético de las estaciones base virtualizadas, manteniendo los requerimientos de calidad de servicio. El objetivo es diseñar un sistema de políticas de energía capaz de adaptarse a las necesidades del usuario y a las condiciones de la red maximizando el rendimiento del sistema. Se han hecho experimentos tanto en escenarios estáticos como dinámicos para calcular el rendimiento de la propuesta en presencia de rápida variabilidad y cambios de repentinos de restricción. |
| SORUS-RAN A3.1-E2 | Second version of the prototype to validate vRAN orchestration | This document corresponds to the deliverable SORUS-RAN-A3.2-E2. This work presents solutions to two problems detected in vRANs caused by resource sharing. The first part introduces AIRIC, a system that provides dynamic reconfiguration based on Reinforcement Learning, addressing the noisy neighbors problem. A 17% reduction in resource utilization has been achieved by correctly sizing the set of computing cores over time. The second part is dedicated to describing the implemented framework to solve the optimal allocation of LLC (L3) cache memory. For this, MemorAI is introduced, which, through a Digital Twin and a neural network-based classifier, enables optimal LLC allocation with energy savings between 0.35 kJ and 1 kJ depending on the cache way allocation strategy. |
| SORUS-RIS-A1.1-E1 | First version of the architecture | The main objective of the three 6G-SORUS projects to study the integration of UAVs with RIS and vRAN. This integration can provide several benefits, including increased coverage and capacity, improved spectral and energy efficiency, and enhanced security and privacy. UAVs can be used as flying base stations to provide wireless coverage to areas where it is difficult or impossible to deploy traditional base stations. By using RIS, the UAVs can enhance the signal quality and coverage, while vRAN can provide dynamic resource allocation and efficient management of the wireless resources. This document describes the architectural elements for the integration of RIS in the operation of the network. |
| SORUS-RIS-A1.2-E1 | RIS Use Cases | This document corresponds to deliverable A1.2 - Use Cases of RIS - which is in the initial phase of the tasks described in the technical specifications. This document presents a preliminary study of the technology and technical feasibility necessary to address the challenge of improving wireless networks through the use of reconfigurable smart surfaces. Three main areas of improvement have been defined: connectivity and reliability, localization and detection, and sustainability and security. It theoretically describes how this technology can be useful to overcome problems in urban environments with high density of obstacles or without direct line of sight from the base stations (BSs). In indoor and outdoor environments, it could be used for people detection as a passive radar or environment mapping. Regarding security, it can enable highly directional communications, resulting in greater energy efficiency and preventing signal interception by other agents. |
| SORUS-RIS-A2.1-E1 | Measurement Platform | This report introduces the environment used to measure the energy consumption of smartphones, based on the architecture of Battery Labs (Varvello, 2022). After describing the environment, our focus will shift to analyzing radio access technologies (3G, 4G, 5G, and Wi-Fi) and their impact on energy consumption. Subsequently, we will outline the planned experiments for the upcoming Deliverable 2. Finally, from Telefónica's perspective as a connectivity provider, we will explore the energy-saving mechanisms of base stations when entering different suspension patterns. These results will be presented in Deliverable 3. |
| SORUS-RIS-A2.1-E2 | Experiment Design and Development | In the field of telecommunications, understanding the energy consumption of smartphones is a critical area of focus. With approximately 8 billion smartphones in use worldwide, examining how these devices consume energy during internet use can highlight opportunities for significant improvements and optimizations. Despite its importance, this area remains relatively unexplored by researchers. Our goal is to identify the main factors contributing to the increased energy consumption in smartphones during internet use. To achieve this, we have concentrated on key elements such as content size, radio access technologies (3G, 4G, 5G, WiFi), device age (newer vs. older models), mobile network operators (e.g., Vodafone, Movistar), and types of applications (web browsers, social networks, video streaming). By identifying these factors, we aim to uncover insights that can drive more efficient energy use in smartphones. |
| SORUS-RIS-A2.2-E1 | Capture methodologies and initial classification of UEs according to the behavioral model | This Deliverable describes the activities carried out during the first period of SORUS-RIS-A2.2. These include the identification of the psychological and behavioural concepts of relevance for the characterization of users’ response to the system latency, as well as the identification of the methodology to be used for its measurement. Furthermore, this Deliverable presents the first version of the user classification algorithm, applied to three different datasets, as well as the results of the pilot study based on the use of psychophysiological signals, as a first step for the integration of these in the classification algorithm. |
| SORUS-RIS-A2.3-E2 | Initial version of the RIS Simulator | This report represents an investigation into the benefits of incorporating RIS (Reconfigurable
Intelligent Surface) equipment into Radio Access Networks (RAN) to enhance cellular services. In the first part of this study, we gauge the benefits of including RIS elements towards radio coverage improvement in real-world scenarios. Specifically, we select a crowded city location with a dense topography and an environment that might account for diversity in the propagation radio models. For this setup, we collect the real-world radio network deployment of a commercial network and derive the theoretical radio coverage based on the network deployment. We then feed this to the RIS-augmented Wireless InSite ray-tracing simulator to account for the strategical deployment of RIS within each specific scenario. Our goal is to demonstrate the impact of RIS deployments in different environments and RIS configurations. Our findings advocate for RIS as a cost-effective
solution for expanding coverage in real-world urban mobile networks.
In the second part of this study, we employ a different ray-tracing tool, namely Sionna RT (by Nvidia). To this end, we develop a modular system-level simulator to analyze and optimize wireless network environments. The simulator includes blocks for BS/user deployment, channel computation, and KPI calculation, and captures key performance indicators such as coverage, SNR, SINR, and data rates. By utilizing datasets from Telefónica’s cellular network, we enhance the simulator to accurately model real-world deployments and integrated Nvidia's Sionna RT for advanced channel modeling, allowing us to assess propagation scenarios with high precision. Key features of Sionna RT, such as RIS placement and reradiation mode optimization, enable us to evaluate the impact of Reconfigurable Intelligent Surfaces on network performance.
|
| SORUS-RIS-A3.1-E1 | RIS Simulator Design | This report represents a preliminary investigation into the benefits of incorporating RIS (Reconfigurable Intelligent Surface) equipment into Radio Access networks (RAN) to enhance cellular services. To conduct this evaluation, it is necessary to design a system-level simulator capable of capturing both realistic propagation environments and the behavior of RIS. In this work, we have designed such a simulator using the Remcom Wireless InSite ray-tracing tool, an advanced commercial software for radio-level analysis. Our goal is to augment the capabilities of this tool to aid in the deployment and simulation of RIS, featuring specific radiation patterns derived from laboratory test bench environments. We aim to assess the impact of RIS repeaters on radio coverage improvement, taking into account key system parameters such as size, location, and orientation. |
| SORUS-RIS-A3.2-E1 | Platform design | This document includes the requirements imposed to achieve a passive, scalable and sustainable RIS. In contrast to other works, a RIS design is proposed with modularity features and capable of offering 3-bit resolution, which enables high spatial resolution codebooks. A RIS design is proposed using an RF switch, which reduces the cost of controlling the reflector cells with respect to PIN diode-based solutions, and the use of two buses called cell selection and phase selection buses, which enable scalability. |
| SORUS-RIS-A3.2-E2 | Operational RIS platform | This document corresponds to the deliverable SORUS-RIS A3.2-E2, which is in the initial phase of the tasks described in the technical specifications. This document explains the practical implementation and empirical characterization of a RIS prototype based on the design presented in the SORUS-RIS A3.2-E1 document, where a system based on reconfigurable intelligent surfaces capable of adjusting the phase of incident radio frequency signals for passive 3D beamforming was introduced. The construction on a printed circuit board and all the design decisions involved in the process are detailed. The prototype is validated through various experiments, including beam patterns, received power measurement, and scalability analysis, as well as a cost study associated with the production of each cell. |
