Key Enabling Mechanisms for Mission-Critical Internet of Things

Publication Type:
Thesis
Issue Date:
2023
Full metadata record
Internet-of-Things (IoT) networks are composed of devices generating varying amounts of data with diverse quality of service (QoS) requirements. Mission-critical IoT networks aim to support applications requiring ultra-reliability and low-latency communication interfaces. Fulfilling the application-specific QoS requirements becomes challenging when network parameters change dynamically in IoT networks having limited radio resources. The efficient use of available time and frequency resources in these networks relies on choosing a network access mechanism, which controls channel access when devices communicate over shared channels. The centralized network access schemes cause additional latency due to the involvement of feedback and control signaling overheads. Therefore, device-level learning-based distributed network access mechanisms are essential for designing wireless networks supporting mission-critical IoT applications. To design such distributed network access schemes, statistical learning and multi-agent multi-armed bandit (MAB) learning are promising tools that address decision-making problems in dynamic environments. This thesis aims to design distributed network access schemes for IoT networks in which massive devices generate delay-sensitive and delay-tolerant data and communicate over shared radio resources. Firstly, we enable the end devices in multi-channel slotted ALOHA-based networks to predict the retransmission limit according to a given latency-reliability criterion. Secondly, we propose a statistical learning-based grant-free network access mechanism for delay-sensitive IoT applications. The proposed mechanism employs a static resource allocation strategy, enabling end devices to use their transmission history to predict different network parameters. Thirdly, to improve the utilization of available radio resources, we design an adaptive network access mechanism operating in a semi-distributed manner. Under this mechanism, we propose a novel grant-free access scheme using a statistical learning approach that enables IoT entities to perform delay-sensitive and delay-tolerant transmissions over dynamically partitioned resources in a prioritized manner. Finally, we propose a multi-agent MAB learning-based grant-free access mechanism for ultra-dense IoT networks, where multiple base stations serve a large number of delay-sensitive and delay-tolerant devices. The proposed mechanism enables the devices to improve their base-station (BS) selection over time to maximize the number of devices connecting to each BS when they meet a prescribed latency-reliability criterion. This thesis demonstrates that the distributed network access approach can efficiently manage delay-sensitive and delay-tolerant transmissions over shared radio resources in a dynamic environment. Moreover, this thesis opens new research directions in designing device-level learning-assisted wireless networks to support heterogeneous IoT applications.
Please use this identifier to cite or link to this item: