The development roadmap for 5G systems and networks is expected to exacerbate trends already visible in current wireless networks: the coexistence of heterogeneous networks over overlapping frequency bands, the availability of multiple Radio Access Technologies (RATs) in each device, and an increasing device and user spatial density.
5G networks will thus pose the problem of how to determine the optimal end-to-end path taking into account all available RATs and their potential mutual impact: current state of the art contributions only analyze it under a limited subset of scenarios in terms of number of RATs, with mutual interaction between them typically not being considered.
The Small World routing In heterogeneous Multi-RAT networks (SWIM) project will tackle the problem of routing in multi-RAT network by proposing a routing strategy, composed by a routing metric that considers Quality of Service indicators, energy efficiency and mutual impact between RATs, and a routing protocol taking advantage of expected 5G RATs features, such as accurate positioning. The routing strategy will select end-to-end paths so to guarantee that the resulting network topology will show the desirable properties of a small world graph, that is high local connectivity combined with short average path length.
The proposed routing strategy will be evaluated by means of computer simulations, carried out in an open source network simulator, extended and tailored so to allow the modeling of 5G multi-RAT networks; the software modules developed during the project will be made available to the research community, and will constitute a valuable output of the project.
The project will also focus on disseminating the research results, by setting up a publication plan and manuscript preparation time line at the beginning of the project, so to ensure a significant impact of the activities on the research community.
The proposed research project is highly innovative, as it tackles a problem that is becoming more and more relevant as the 5G roadmap proceeds, and for which no standardized and agreed upon solution exists: how to determine the best end-to-end multihop path across multi-RAT devices with different capabilities and experiencing different environmental conditions. The project will indeed go beyond the current state of the art under several aspects, listed below:
1) definition of models for interaction between RATs to be adopted in the routing metric ¿ given the wide number of RATs to be available in 5G networks, it can be expected that many of them will share the same frequency ranges, in particular in the unlicensed bands. The proposed metric will take into account the mutual impact of the RATs, and its definition will thus require to model such impact, something that is not currently addressed in routing proposals for multi-RAT networks, where the different RATs are considered to be orthogonal in terms of physical resources.
2) inclusion of topological aspects in route selection ¿ the topological parameters relevant to the small world behaviour will be taken into account in the route selection, so to maximize the probability for this behaviour to emerge. As an example, in a greedy algorithm with hop-by-hop route selection, the probability of selecting a long distance RAT would increase as the selected route hop count increases, in order to limit the total number of hops.
3) Inclusion of current active link topology in route selection ¿ a key aspect in ensuring that a small world behaviour emerges in the network is to take advantage as much as possible of existing shortcuts before activating new ones: for this reason existing links will be taken into account in the setup of a new path, and used whenever possible, compatibly with the QoS requirements of the new routing request: this aspect is often neglected in existing approaches, that consider each request independently.
4) integration of the wired network portion in the routing problem ¿ the 5G environment is expected to be populated by a combination of Device-To Device (D2D) technologies and access technologies connecting devices to the wired network: the routing strategy will include potential ¿shortcuts¿ provided by such wired network in the definition of the routing path. This will allow to consider a wide range of intermediate network scenarios lying in between the two main scenarios considered so far in routing for wireless networks: a ¿sink¿ scenario where the goal is to send data through wireless links to a sink providing connectivity to the wired network, and an ¿adhoc¿ scenario where the wireless network is completely isolated from the wired network.
5) Cross-RAT routing optimizations ¿ the routing protocol to be designed jointly with the routing metric will take advantage of the possibilities made available by each RAT, irrespectively of the actually selected RAT. As an example, if a RAT offers accurate positioning of neighboring devices, this information will be used in routing even if the RAT selected for the next hop does not offer this functionality.
6) Analysis of network scalability in realistic scenarios ¿ network scalability was discussed in the past for purely ad hoc, homogeneous networks; this project will assess it in heterogeneous networks based on 5G use cases and scenarios.
7) Definition and release of software modules implementing multi-RAT routing ¿ currently, open source simulators only provide a basic support to multi-RAT networks; vertical handover (i.e. handover between different RATs) is only partially implemented, and multihop routing only considers a single RAT. As a result of project activities, modules implementing multi-RAT routing will be written and integrated in the selected open source simulator, and eventually released to the research community, thus increasing the impact of the project.