Principal investigators: Marco Caccamo (TUM), Raymond KNOPP (IMT), David Gesbert (IMT); Other contact person: Harald Bayerlein (TUM), Omid Esrafilian (IMT)

Short summary and central question:

As 5G networks are being rolled out, many studies and efforts have begun exploring the next generation of wireless networks (6G). It appears that 5G in its current standardized form will fall short of fulfilling all expectations in the face of extraordinary future challenges. The main principles of future mobile communication systems are the ability to handle a higher degree of flexibility and functionality which enables future intelligent connectivity and control along with joint communication and sensing; Moving away from traditional network architectures with passive terminal devices towards ultra-flexible and self-configurable network architectures with intelligent and connected devices. By capitalizing on the internode connectivity provided by networks, the intelligent nodes can also perform collaborative sensing and control. 

Having an ultra-dynamic network becomes extremely important when it comes to fast responses in the face of crises triggered by natural or human-caused disasters such as earthquakes, fires, floods, wars, etc. In such scenarios, the aid of a connected dynamic network comprising intelligent nodes, such as UAVs and ground mobile robots, can save lives and help society resume its normal operation quickly. For instance, UAV-equipped cellular access points can establish a temporary network allowing radio access to users and at the same time, they can provide sensing and localization services which is of crucial importance in search and rescue scenarios. Beyond the crisis scenario, intelligent and connected networks can also boost efficiency in the ICT sector under normal circumstances. Collaborative learning and sensing are essential requirements of future smart cities which inherently come with new challenges. For instance, learning often suffers from the heterogeneity of devices in terms of computation capability and the richness of training datasets. An intelligent network can coordinate between the nodes to compensate for this heterogeneity to harmonize the 4 learning process for all nodes by optimally exchanging the data among them. Another example is self-driving cars. When it comes to autonomous vehicles, having a sufficient understanding of the environment is crucial. An intelligent network, which comprises several smart nodes including aerial access points and smart cars, can optimally exchange the data between all the cars to provide a global understanding for each of them and at the same time can improve the localization of the cars on the ground with the help of the aerial nodes. In other words, intelligent networks not only can guarantee seamless connectivity between the users but also can provide them with several other services such as accurate sensing and localization which can be a reliable alternative to the Global Positioning System (GPS). 

Joint control, sensing, and communication for connected intelligence

 Inspired by the examples above, this project proposes to study the design and applications of connected and intelligent networks. Given the anticipated transition to intelligent networks, the implications of this transition are expected to be in network management. Most of the management and control tasks in current networks are based on centralized decisionmaking. This makes them unsuitable to be used in future intelligent networks. In fact, what is needed is a significant redesign and reconsideration of the fundamentals and principles of distributed and collaborative network management. In particular, this project will look at the problems of cooperative sensing and control of the smart nodes in the network to provide seamless connectivity between the users as well as to offer other services such as localization when needed. We also investigate the problem of decentralized and collaborative learning between the nodes first to reduce the signaling overhead between the learners and at the same time to address the heterogeneity of the nodes. Moreover, we will lay the groundwork for developing a prototype for future collaborations. This prototype consists of several intelligent nodes/robots (e.g. UAVs, rovers) equipped with radio modules that are based on OpenAirInterface (OAI) [1] for 5G developed by EURECOM.

Overview of the state-of-the-art 

Traditionally, control, sensing, and communications have been looked at separately. While the problem of controlling autonomous robots, e.g. navigation or surveillance using drones, has been studied in the robotics community [2-4], the possibilities to integrate cellularconnected robots (e.g. UAVs) attached to mobile network links or robots/UAVs providing communication services in support of terrestrial networks have been investigated separately by researchers in communications [5-7]. Moreover, in the context of 6G research, there has started a convergence between sensing and communications [7, 8], where radio nodes can also assist with sensing and localization. Contrary to static or uncontrolled mobile devices in the network, we can optimize node placement to improve the sensing and localization performance [9]. The nodes can also collaborate for further performance improvement. 

A central challenge that this project will consider is to provide a collaborative and decentralized network management framework. Within this framework, each node in the network acts in a way not only to successfully perform its local task but also to contribute to the improvement of the global network performance. For instance, UAVs as flying access points in the network can optimize their locations under flying energy constraints to maintain some level of connectivity among ground users and at the same time improve the localization service provided to some of the ground nodes.

Objectives of the project 

–  Capitalizing on the insights and the expertise of both teams, we aim to build a new collaboration identifying the central challenges associated with future 6G connected intelligence.

 – Define specific research questions that can form the basis for a larger funding proposal and seek joint funding from third-party agencies and industries.

 – Develop algorithms and simulation environments for collaborative decentralized decision-making in mesh networks of autonomous agents balancing local goals and network connectivity with a focus on reducing training data demand for increased energy efficiency that are suitable for practical deployments.

 – Develop joint experiments based on EURECOM’s Drone4Wireless testbed lab with a focus on disaster response scenarios, i.e. joint communications and localization of network users in peril. The prototypes developed under this lab won the “Fundamental Research Project of the Year” awarded by the French SCS (Secured Communications) Research and Industry Cluster (Pole Competitivité), 2019.

 – Publish results in venues and journals of the robotics and communications research communities emphasizing the interdisciplinary character of the 6G connected intelligence challenges and fostering dialogue.

Expected impact on academia, industry and society 

Societal and industrial impact 

The recent public health crisis, political tensions and the European security situation, and ever-looming climate change all threaten the stability of international cooperation and trade. In the face of these events, it is imperative that Europe intensifies efforts in research relevant to critical technologies like future 6G connected intelligence. It seems likely at the same time that previous reservations about the progression of industrial automation will dissipate and industry transformation will accelerate, making the investigation of joint control, sensing, and communication problems in this context a top priority. 

Research and academia

 With the integration of autonomously acting agents into future communication networks, fundamentally new challenges in decision-making arise and the development of flexible MLbased algorithms becomes necessary. By highlighting the connection between the two mostly disjoint research areas of wireless communications and real-time embedded solutions for robotics and bringing them together through this project, we expect to harness the synergies that exist between these two communities of researchers, finding better overall solutions to challenges that exist in both research fields. From an academic point of view, we will publish results in venues and journals of both research communities to foster dialogue between robotic and communication research communities.