Work package WP4 – Radio resource and operation management
This WP deals with subnetwork operations at higher layers than the ones studied in WP3. The main objectives of WP4 are:
- Study the impact of interference in the identified scenarios and use cases.
- Develop novel mechanisms for 6G in-X subnetworks radio resource management for the case of centralized, distributed, and hybrid management.
- Develop novel mechanisms for detecting and mitigating the impact of jammers and malicious interferers in subnetworks.
- Develop solutions for managing traffic and control/user roles in subnetworks belonging to the same entity.
- Develop solutions for efficient and dynamic traffic and computational offload between subnetworks and the larger 6G network.
Task 4.1 aims to develop radio resource management techniques for in-X subnetworks both when they are in 6G network
coverage (4.1a), when they are out of 6G network coverage or with partial or intermittent connectivity to it (4.1b).
Common for these subtasks is that they will also consider how interferers are identified and how these characteristics can
be exploited by effective radio resource management. Also, techniques for detecting malicious interferers (discriminating
them from legitimate interferers) and mitigate proactively their effects are to be designed.
Subtask 4.1a Centralized radio resource management.
This subtask focuses on the development of new methods for centrally managing radio resources in subnetworks, for the
sake of achieving their performance requirements despite potentially large interference levels. It reflects a scenario where
a broader 6G network with radio control capabilities can manage radio resources of subnetworks in its coverage area, and
instruct them on the transmission mode (i.e., transmit power, frequency resources, precoder) to be used. Such centralized
schemes require local measurements at each subnetwork (performed by the AP, and eventually, by the devices) and
signaling between the AP and the umbrella 6G network. Given the complexity of scenarios and interference conditions,
characterized by large and potentially rapid interference fluctuations, AI methods are expected to outperform heuristics.
AAU and COGN will explore techniques such as Bayesian reinforcement learning and graph neural network methods
for transmission mode selection in dense mobile subnetworks, considering practical limitations due to limited
communication resources between subnetwork AP and 6G umbrella network, as well as signaling delays. The aim is
also to limit as much as possible the sensing operations of the subnetwork devices. Also, we will study the benefits of
introducing context information (e.g., position and heading of subnetworks) in the learning model, for the sake of further
improving spectral efficiency performance.
Bosch will work on novel concepts for dynamically adjusting and aligning the radio resources to be used by different
subnetworks in a certain area, exploiting particularly radio measurements, but also relevant context information (e.g.,
maps) and other sensor data (e.g., cameras, radar, position of a device) that allow for an accurate prediction of the
future interference situation. In addition to that, Bosch will explore to what extent also the behavior of the applications
running over different (adjacent) subnetworks can be aligned and optimized to reduce interference issues, for example
by introducing proper timing offsets in control loops running with the same cycle time.
Related TC: TC12.
Subtask 4.1b: Distributed and hybrid radio resource management
This subtask will seek to develop a distributed interference management scheme allowing the in-X subnetworks to
autonomously manage their radio resources and avoid severe mutual interference. Also, hybrid solutions are to be studied,
where connection to the larger 6G network can be leveraged when present, for the sake of enhancing resource utilization.
AAU will design fully distributed solutions where each subnetwork learns independently the transmission mode for
its devices based on local sensing. Bayesian reinforcement learning approaches will be exploited at this purpose. The
Bayesian model will include in the prior non-idealities arising from noisy and delayed sensing. For hybrid resource
management, AAU will also investigate approaches where a model is centrally trained (e.g., based on graph neural
networks) and then distributed across the subnetworks, which maintain a local version to be used for performing decisions
when the connection with the central controller is intermittent or lost. The performance of the hybrid scheme as a function
of the connectivity quality with the central controller will be studied.
Nokia will study which type of local measurement and cooperation techniques are needed for ensuring efficient resource
allocation in distributed and hybrid conditions. Further, Nokia will study mechanisms to handle the situation where
subnetworks transition from being in coverage to being out-of-coverage, causing a mix of nearby subnetworks that are
in coverage and others that are not.
Related TC: TC13.
Subtask 4.1c: Novel techniques for detection/mitigation of jammers
Besides legitimate interference, subnetworks must be able to counteract malicious interferers, such as jammers. This
subtask will focus on the design of new techniques for detecting jammers (distinguishing them from legitimate
interferers) and mitigating their effects that can harm the support of critical services. These solutions will complement the
jamming robust PHY design by providing a further tier of protection to external attacks. We aim at designing solutions
that minimize the delay between action of the jammers and system response, as well as proactive solutions, where next
actions of the jammer can be predicted and counteracted beforehand. AAU will investigate Bayesian reinforcement
learning approaches for this subtask, that include in the learning phase a-priori knowledge of possible attacker strategies.
Also, rewards will take into account metrics associated to proactiveness of the system. IDE will investigate autoencoderbased
models targeted for jamming detection.
Related TC: TC14.
This task will investigate new techniques for managing traffic and radio resources among subnetworks belonging to
the same entity and their interactions with the outer 6G network. Initially, the routing of data and control signaling
will be investigated for subnetworks in the same entity (e.g., the same vehicle, or the same classroom for consumer
type of applications) and the possibility to coordinate their operation (subtask 4.2a). Subsequently, offloading data and
computation to the edge nodes of the overlay 6G network will be studied, thus enabling the seamless integration data
as well as AI/ML services to the subnetworks (subtask 4.2b). Finally, the case where similar bands are utilized by the
subnetwork and the overlay 6G network will be considered. For the sake of ensuring the seamless operation of both
networks, dynamic spectrum sharing techniques will be developed (subtask 4.2c).
Subtask 4.2a: Routing of data and control signaling within subnetworks in the same entity (e.g., initialization, maintenance, and path redundancy).
Apple will investigate coordination methods between APs of different subnetworks. In order to improve the reliability,
different nodes of the subnetwork can have different and flexible roles in coordinating and combining the communication
links. Furthermore, the mobility of the nodes and the subnetworks will be considered. Also, this subtask will investigate
the distribution of control plane and data plane functionalities among nodes and subnetworks.
UMH will contribute to the design of adaptive and efficient coordination mechanisms within and across subnetworks
and will design networking solutions (e.g., programmable and adaptive redundancy and multi-path links) for dependable
service level provisioning on end-to-end in-X connectivity.
Related TC: TC16.
Subtask 4.2b: Dynamic computational resources offloading from subnetwork AP to 6G edge-cloud
The 6G in-X subnetwork can, when in coverage of a 6G network, be aided by the 6G network to free up some of the
critical computational resources in the 6network by offloading non-latency critical computations to the 6G edge-cloud.
This task will seek to develop seamless offloading to a 6G edge-cloud for both data services, but also for executing AIbased
services, e.g., training of interference or jammers detection models.
Apple will investigate the trade-offs and KPIs for decision making policies for computing distribution. Apple will also
investigate the impact of AI offloading services on the control plane and data plane.
UMH will design mechanisms for E2E service level provisioning considering connections within subnetworks, between
subnetworks, and to the wider area 6G network. Special focus will be placed on deterministic networking mechanisms
within the 6G network of networks. The native integration of in-X subnetworks in the 6G ‘network of networks’ will be
exploited to opportunistically and elastically offload in-X functions to the edge and cloud without compromising service
provisioning in a connectivity continuum subnetwork-edge-cloud with mechanisms to manage (and, when possible,
predict) variations in QoS levels.
IDE will develop a mobile computation offloading platform architecture. An offloading architecture framework will be
developed that enables a mechanism for task offloading between network entities and devices over the wireless channels,
as well as adaptive procedures for entities and devices (including CPU resources) including wireless transmission
configuration. The platform will comprise a code partitioner block, an executioner block and an offloading block for the
remote execution of selected tasks at the remote entity.
Related TC: TC15
Subtask 4.2c: Dynamic spectrum sharing between 6G and in-X subnetwork
When the in-X subnetwork operates at frequency bands close to other 6G networks one option to manage time-variant
radio resource needs (predictable for the in-X subnetwork), is to consider spectrum sharing mechanisms. This subtask
will seek to develop mechanisms for the in-X subnetwork along with the 6G network when they operate at the same or
overlapping frequency bands. It will focus on the frequency and time domain but may also consider the spatial domain,
i.e., coding layers.
Bosch will design and evaluate suitable spectrum sharing mechanisms, e.g., by dynamically assigning certain spectrum
resources to a 6G subnetwork by the corresponding parent network. In this respect, also relevant regulatory constraints
in different countries will be analyzed and considered and a thorough comparison will be made between the pros and
cons of using licensed and license-exempt spectrum for subnetworks.
Sony intends to investigate the possibility of having a unified approach of how a 6G subnetwork can get access to
spectrum resources depending on local regulatory requirements.
Related TC: TC16.