Autonomous driving technologies have attracted much attention to realize smart mobility societies. Autonomous driving strictly requires uninterrupted mobile communications for remote monitoring in order to ensure safe operation. This requirement is hard to achieve with particular technology such as multipath transmission. In this paper, we propose proactive and reactive multipath transmission mechanisms with multilayer bandwidth prediction to find degradation of communication quality and avoid it completely. We implemented the proposed mechanisms into the Cooperative Infrastructure Platform and conducted a field evaluation by driving a vehicle on public roads. Our evaluation results show that a$74-100 \%$recall rate can be achieved in bandwidth prediction. The Cooperative Infrastructure Platform provides stable communication by accurately controlling multiple mobile networks on the basis of the bandwidth prediction, resulting in only total 3 seconds of video stall for$\mathbf{1}$hour of driving.

One of the useful methods of ensuring stable communication for autonomous vehicles is to predict communication quality using radio environment maps. We propose a Kriging method for constructing radio environment maps to improve interpolation accuracy by using anisotropy of the shadowing correlation on the distance and azimuth-angle plane based on the base-station position. Using $2.2-\mathrm{GHz}$ band path-loss data measured in an urban area, we evaluated the accuracy when interpolating the total measurement data of approximately 14,000 points from sample data of 10 and 100 points. We compared the proposed method with four conventional methods: nearest neighbor interpolation, K-nearest neighbors, inverse distance weighting, and conventional ordinary kriging. The median and minimum root means square errors when sample data were extracted 30 times were approximately 8.2 and 6.4 dB when the number of samples was 10, and approximately 4.7 and 4.2 dB when the number of samples was 100. These results indicate that the proposed method can interpolate with higher accuracy than the conventional methods.

In this paper, a system for predicting degradation of wireless-communication quality in the case of autonomous vehicles is proposed. It uses historical-data-based throughput estimation using a neural network trained on location information and past performance. A method for improving the accuracy by fine tuning is also proposed. We evaluate the accuracy of throughput estimation using the proposed system in two outdoor use cases (field and urban) and one indoor use case. The results show that the proposed method of fine tuning using a large amount of received power data and a small amount of throughput data can improve the estimation accuracy of throughput.

Realizing extreme reliability for Internet of Things (IoT) communications is one of the major milestones paving the way towards Beyond 5G (B5G) and 6G. In this work, we investigate the issue of improving the reliability of packet transmissions in the absence of prior knowledge of network statistics, nor of instantaneous Channel State Information (CSI), for B5G Sub-6GHz/mmWave integrated networks. Specifically, the aim is to maximize the global successful packet reception at devices, while guaranteeing their individual Packet Loss Rate (PLR) requirements. The proposed method exploits a newly developed approach of Risk-Averse Reinforcement Learning (RARL), for exploiting multi-connectivity over Sub-6Hz and mmWave interfaces. Namely, the Access Point (AP) is able to optimize its interface selection decisions despite the unknown dynamics of the wireless environment based on limited feedback from its associated devices, so as to increase reliability under low delay and resource consumption. Numerical results show that, the proposed method significantly improves the global reliability performance by rapidly learning and adapting its decisions as compared to baseline methods.

In this work, we investigate the problem of joint user-to-access points (AP) association and beamforming in an integrated sub-6GHz/mmWave system. The goal is to maximize the long-term throughput of the system, while satisfying a large number of heterogeneous user QoS requirements in a distributed manner. We propose a method based on Deep Q-Networks (DQN), where each user self-optimizes its AP association and interface requests, and can be served by several APs simultaneously for supporting multiple applications. Based on these requests, each AP selects its associated users and applications served on each interface, while optimizing its mm Wave beamforming parameters. Simulation results show that, compared to baseline DQN schemes among which the Action Elimination (AE)-DQN, the proposed method enables to fine-tune the selection of APs and interfaces to the specific level of each required QoS, thereby achieving a high global throughput while notably reducing user outage probabilities1.1.This collaborative research project is funded by NTT Corporation, Japan.

This work investigates the design of a sustainable Deep Q-Network (DQN) implemented at the user device, whose purpose is to optimize the user’s association to multiple access points (AP) in a Beyond 5G (B5G) Sub-6GHz and mmWave integrated network. To better cope with dynamic mobile environments, we first propose an adaptive $\varepsilon$-greedy policy at each user’s DQN in order to maximize the long-term sum-rate while simultaneously satisfying the Quality of Service (QoS) constraints of different applications. We then provide the detailed analysis of the energy consumed by each user device, in particular the power for DQN processing and for data movement. The trade-off between network performance in terms of sum-rate and QoS outage probability, and energy consumption at the user side is evaluated. Numerical results not only show the effectiveness of the proposed method compared to baseline, but also reveal the tremendous energy costs required by the default user DQN, underscoring the paramount importance of the proposed trade-off aware user DQN design1.

Future wireless networks will be facing unprecedented difficulties arising from mobile traffic growth, network densification, as well as diversification of applications and services. Indeed, future user devices are expected to integrate diverse radio interfaces such as 5G, WBAN or IoT, enabling each user to be served a wide range of applications at any time. This poses significant challenges in terms of wireless resource sharing and interference management, as more and more stringent Quality of Service (QoS) constraints should be jointly satisfied in dense interfering environments. Furthermore, future networks are expected to be highly autonomous and decentralized. To meet these challenges, this work proposes distributed user-to-multiple Access Points (AP) association methods, where the objective is to maximize the long-term sum-rate subject to application QoS constraints, as well as to AP load constraints. Our distributed methods enable each user to leverage their Deep Reinforcement Learning (DRL) capabilities, in particular Deep Q-Learning (DQL), to self-optimize their APs’ selection solely based on their local network state knowledge, so as to best satisfy their diverse requirements. Numerical results show that, compared to baseline schemes, the proposed methods enable global throughput enhancements while reducing user QoS outage probabilities, even in large and dense networks.

This paper describes a novel wireless resource allocation method we have developed with distributed architecture and the DB search algorithm. Computer simulation results revealed that the method improves throughput 1.2 times on average compared to the conventional fixed allocation method with centralized architecture. It also reduces the computational complexity by three digits compared to the conventional metaheuristic method. Our method will make it possible to achieve large scalability of wireless resource optimization on future wireless access systems, such as multi-RAT, multi-cell environments of 5G and 6G, or high density wireless LAN systems.

This paper describes cooperative wireless LAN (WLAN) control methods we propose for high density environments, with the focus on IEEE 802.11ax WLAN systems. It also shows how they were evaluated by computer simulation. Two control schemes are proposed, the first of which optimizes spatial reuse parameters and access network selection. This scheme improves user equipment throughput 1.7 - 2.4 times in terms of throughput distribution (median value). The second scheme is for cooperatively scheduling trigger frames with hierarchical clustering. This scheme enables capacity to be improved by up to 11.3% when WLAN access points are deployed closely. These control schemes are potentially effective for IEEE 802.11ax WLAN systems utilized as 5G mobile network components.

Throughput starvation, which occurs due to the well-known exposed/hidden terminal problem in CSMA/CA based channel access mechanisms, will be a critical problem for dense wireless LAN systems. To address this problem, we previously proposed schemes that identify and manage starved devices by monitoring the beacon signals transmitted by access points (APs) and showed that they can identify low throughput devices and have potential to improve WLAN system throughput. This paper describes experimental results obtained for schemes that identify starved APs and manage their channels. Numerous experiments verified that the schemes can identify APs that have quite low throughput as starved APs and that management can reduce the low throughput APs in real environments. We also indicate that it is necessary to consider beacon transmission delay fluctuation when determining the threshold used for identifying starved APs in cases such as the surrounding environment fluctuating in a short period.

Data traffic has been increasing rapidly in recent years on mobile networks. Several methods using unlicensed bands for LTE are proposed to increase wireless capacity. Stand-alone unlicensed LTE is also proposed as a technology that provides LTE service with only unlicensed bands. In this paper, we propose distributed P-GW systems for stand-alone unlicensed LTE base stations that can provide future mobile services by using an ad-hoc network architecture. Additionally, we evaluate the effectiveness of our proposal in terms of network load, handover latency, and user data plane latency in comparison with the full Cloud EPC model and IPsec model. We reveal that the proposed architecture has an advantage in terms of network loads and end-to-end latency and has an issue with handover latency.

Uploading of huge files to cloud servers through mobile communication has been becoming more popular in recent years. However, current uploading throughput is frequently limited due to bottlenecks in server processing ability and/or narrow bandwidth sections of networks. For this study, we aimed to improve the user experience of uploading in future wireless access environments. We propose a novel upload cache architecture that flexibly uses computer resources located on the network edge and achieves high-speed uploading with parallel uploading of divided files. We also report on the evaluation results of throughput improvement (more than 10 times faster) by implementing our proposed architecture to prototype software.

Multi-hop communication systems have been widely studied and many of them are based on the IEEE 802.11 WLAN system. The paper explains the problem of throughput degradation of a relay station when IEEE 802.11 DCF is adopted in multi-hop communications. The relay stations lose some of their capacity compared to stations that are not relaying the data of other stations. Because all stations, under the CSMA/CA protocol, have equal transmission opportunity, the relay stations have to send packets from other terminals. We propose an autonomous priority control method that changes the contention window size depending on the number of relay stations and the number of hops to ensure that relay stations are offered the same transmission capacity as regular stations in the service area. Then its performance is evaluated by computer simulations.