The growing demand for highly reliable, low-latency communication in industrial and real-time applications poses significant challenges for traditional Wi-Fi networks, especially under dynamic traffic conditions. Multi-Connectivity (MC) presents a viable solution to enhance network performance by utilizing multiple transmission paths. This paper extends our previous work where we proposed Channel Capacity Index-Aware Adaptive Scheduler (CIAS), and evaluates static and adaptive scheduling, i.e., selecting which paths are used, on latency and reliability in non-periodic, sporadic, and attenuated scenarios, which are commonly observed in autonomous systems and industrial applications. We analyze the performance of static schedulers and compare them with CIAS, an adaptive cross-layer scheduling approach that dynamically selects the optimal MC scheme based on channel capacity. Our results demonstrate that CIAS consistently reduces latency and mitigates packet delays, particularly under high traffic loads, sporadic transmissions, and signal attenuation. Furthermore, CIAS enhances reliability by minimizing tail-end latencies and adapting to channel fluctuations more effectively than static schedulers. This study underscores the critical role of adaptive MC scheduling in improving wireless communication efficiency for time-sensitive applications that demand both low latency and high reliability.

The upcoming Wi-Fi 8 standard is heading towards reliability, with a strong focus on machine-type communications and time-critical applications. In this work, we present a novel approach to ensure the Quality of Service (QoS) in Wi-Fi systems by leveraging Machine Learning (ML) models for proactive channel switching. We propose two time-series prediction models in this regard. The predictive QoS (pQoS) model assesses the probability of exceeding a packet error rate (PER) threshold in the near future, enabling proactive channel switching decisions during operation. To make an informed decision on which channel to switch to, the Channel Quality List (CQL) model classifies candidate Wi-Fi channels in terms of the expected PER based on sensed radio activity.We evaluate this ML-aided proactive channel switching procedure through simulation of a mobile robotics use case in different interference scenarios. To generate realistic training data, a statistical interferer model has been developed, which generates dynamic traffic patterns and user behavior based on real-world measurements. This model captures the variability in possible interference scenarios, providing a robust foundation for training the ML models, and for evaluating the trained models during simulation.The proposed ML-aided channel switching approach allows the system to quickly switch away from a deteriorating channel quality before the QoS suffers, and to switch to the best channel in terms of PER. The prediction models improve the QoS considerably, where the Packet Error Rate (PER) is reduced by up to 99 % of the PER of a non-switching system in the best cases, and by 95 % even in very difficult scenarios. This work highlights the substantial benefits of employing ML for predictive channel switching in Wi-Fi systems, offering a robust solution to the growing challenge of interference in congested radio environments.

Measurements of one-way delay in 5G networks are essential for evaluating system performance in various practical scenarios. This paper presents a comprehensive study comparing three different packet generators on real-world deployments and their associated challenges. We explore different topolo-gies, types of terminals, and network congestion conditions, discussing the suitability and limitations of different packet generators. Moreover, we propose practical considerations to overcome limitations in capturing dynamic and heterogeneous 5G network environments and provide insights for system designers in precisely measuring and characterizing one-way delay in 5G networks for real-world scenarios.

IEEE 802.11ax, also named Wi-Fi 6, introduces significant enhancements to the channel access mechanism. The introduction of orthogonal frequency-division multiple access (OFDMA) and the more centralized channel access coordinated by the AP can potentially improve the latency and reliability performance of Wi-Fi 6 compared to previous Wi-Fi generations. This would make Wi-Fi systems suitable also for more demanding applications, e.g., in the industrial domain, which require low communication latency with high reliability. However, only few works exist that evaluate these new features with regard to such applications. This work presents a comprehensive measurement study on the latency and reliability performance of Wi-Fi 6 and compares it to performance of Wi-Fi 5. The baseline scenario measurements reveal a significant latency reduction of Wi-Fi 6 compared to Wi-Fi 5 especially for larger number of stations and especially in uplink direction. Since Wi-Fi 6 operates in unlicensed bands like its predecessors, the influence of interfering networks on the communication performance is an important aspect. Our evaluation shows that interfering traffic can significantly increase the latency and packet loss experienced by Wi-Fi 6 for smaller operation bandwidths, while for larger bandwidth of 80 MHz and resulting lower relative network load Wi-Fi 6 can still provide 99-percentile latency below 10 ms in the evaluated scenario. Overall, the achieved results with Wi-Fi 6 are promising, but further evaluations especially on the effect of different OFDMA scheduling algorithms and configurations is needed to evaluate its potential for demanding applications.

The demand for wireless communication systems that provide low latency with high reliability is growing due to emerging applications such as wireless industrial control or mobile robotics. A promising approach to improve latency and reliability performance of wireless communications is Multi-Connectivity (MC), i.e., using multiple communication paths at the same time. The task group working on the upcoming 802.11be amendment is discussing multi-link operation as one of the main features. In previous work different static MC scheduling schemes have been evaluated and in different scenarios different schemes demonstrated to provide the best latency and reliability performance. Based on these findings and in coherence with the discussion about multi-link operation in the 802.11be task group an adaptive MC scheduling scheme is proposed in this work. Two variants are proposed and both perform a client-based decision. The first variant decides solely based on the parameters available at every station itself. The second variant also takes into account additional information about the network that are gathered and shared with the stations by the AP. System level simulations with ns-3 for an industrial scenario show that the adaptive MC schemes can significantly improve the latency and reliability performance, especially in scenarios with medium network load. The insights gained in this work can be used to develop multi-link operation concepts in order to improve latency and reliability performance in the upcoming 802.11be amendment.

Wireless communication systems which provide low latency with high reliability are required for emerging industrial applications such as closed-loop control systems or mobile robotics. Multi-Connectivity (MC), i.e., using multiple communication paths simultaneously, can be a measure to enhance latency and reliability performance of wireless communication systems. MC schemes can be applied on different layers of the communication stack. While Physical (PHY) Layer MC schemes combine the paths at signal or symbol level, in higher layer schemes, i.e., MC schemes on MAC Layer or above, the paths are combined at packet level. Higher layer schemes provide more degrees of freedom for dynamic scheme adaptation, while PHY MC schemes have more information available. MC schemes on different layers need to be compared to evaluate their suitability to enhance latency and reliability performance in different scenarios. In this work system level simulations are performed with the network simulator ns-3 to compare PHY Layer MC schemes, i.e., selection combining (SC), maximum ratio combining (MRC) and joint decoding (JD), with higher layer MC schemes, i.e., MP-UDP with packet duplication (PD). Our results suggest that the gain of PHY Layer MC schemes increases with decreasing SINR. The higher layer MC schemes provide diversity not only on PHY but also on MAC and higher layers and could thus decrease access and queuing latency. The gain of MP-UDP increases with increasing network load. The insights obtained in this work can be used to develop dynamic MC schedulers that activate or deactivate certain MC schemes or switch between MC schemes on different layers based on the scenario.

Emerging applications such as wireless industrial control or robotics raise strict requirements in terms of latency and reliability on wireless communication systems. It is still an open issue which wireless communication system can be used to enable industrial applications such as closed-loop control. A promising approach to improve latency and reliability of wireless communications is multi-connectivity (MC), i.e. using multiple communication paths simultaneously. Different scheduling schemes can be used to distribute the traffic over multiple links. The characteristics of these schemes and which one is best suited to enable reliable low-latency communications in different scenarios needs to be investigated. In this paper, we evaluate the latency and reliability performance of a local Wi-Fi and a private LTE network for traffic patterns as envisioned for industrial applications. Moreover, we assess the performance of different MC scheduling schemes operating on Application Layer over these two wireless links with focus on reliability and latency metrics. For the evaluated single-user scenario, WiFi provides a lower mean latency than LTE. The evaluation of MC scheduling schemes shows that packet duplication (PD) stabilizes the latency by mitigating outliers, while load balancing (LB) reduces the latency of nearly 50 percent of packets in a scenario with bad radio conditions. Our results suggest that using links with similar mean latency would be beneficial for all scheduling schemes.