https://hdl.handle.net/10388/14431

Rahman, Gazi Mohammad Ali Ehsan ur (2023) An Internet of Things (IoT) based wide-area Wireless Sensor Network (WSN) platform with mobility support., USASK Harvest - Theses and Dissertations, https://hdl.handle.net/10388/14431

Wide-area remote monitoring applications use cellular networks or satellite links to transfer sensor data to the central storage. Remote monitoring applications uses Wireless Sensor Networks (WSNs) to accommodate more Sensor Nodes (SNs) and for better management. Internet of Things (IoT) network connects the WSN with the data storage and other application specific services using the existing internet infrastructure. Both cellular networks, such as the Narrow-Band IoT (NB-IoT), and satellite links will not be suitable for point-to-point connections of the SNs due to their lack of coverage, high cost, and energy requirement. Low Power Wireless Area Network (LPWAN) is used to interconnect all the SNs and accumulate the data to a single point, called Gateway, before sending it to the IoT network. WSN implements clustering of the SNs to increase the network coverage and utilizes multiple wireless links between the repeater nodes (called hops) to reach the gateway at a longer distance. Clustered WSN can cover up to a few km using the LPWAN technologies such as Zigbee using multiple hops. Each Zigbee link can be from 200 m to 500 m long. Other LPWAN technologies, such as LoRa, can facilitate an extended range from 1km to 15km. However, the LoRa will not be suitable for the clustered WSN due to its long Time on Air (TOA) which will introduce data transmission delay and become severe with the increase of hop count. Besides, a sensor node will need to increase the antenna height to achieve the long-range benefit of Lora using a single link (hop) instead of using multiple hops to cover the same range. With the increased WSN coverage area, remote monitoring applications such as smart farming may require mobile sensor nodes. This research focuses on the challenges to overcome LoRa’s limitations (long TOA and antenna height) and accommodation of mobility in a high-density and wide-area WSN for future remote monitoring applications. Hence, this research proposes lightweight communication protocols and networking algorithms using LoRa to achieve mobility, energy efficiency and wider coverage of up to a few hundred km for the WSN. This thesis is divided into four parts. It presents two data transmission protocols for LoRa to achieve a higher data rate and wider network coverage, one networking algorithm for wide-area WSN and a channel synchronization algorithm to improve the data rate of LoRa links. Part one presents a lightweight data transmission protocol for LoRa using a mobile data accumulator (called data sink) to increase the monitoring coverage area and data transmission energy efficiency. The proposed Lightweight Dynamic Auto Reconfigurable Protocol (LDAP) utilizes direct or single hop to transmit data from the SNs using one of them as the repeater node. Wide-area remote monitoring applications such as Water Quality Monitoring (WQM) can acquire data from geographically distributed water resources using LDAP, and a mobile Data Sink (DS) mounted on an Unmanned Aerial Vehicle (UAV). The proposed LDAP can acquire data from a minimum of 147 SNs covering 128 km in one direction reducing the DS requirement down to 5% comparing other WSNs using Zigbee for the same coverage area with static DS. Applications like smart farming and environmental monitoring may require mobile sensor nodes (SN) and data sinks (DS). The WSNs for these applications will require real-time network management algorithms and routing protocols for the dynamic WSN with mobility that is not feasible using static WSN technologies. This part proposes a lightweight clustering algorithm for the dynamic WSN (with mobility) utilizing the proposed LDAP to form clusters in real-time during the data accumulation by the mobile DS. The proposed Lightweight Dynamic Clustering Algorithm (LDCA) can form real-time clusters consisting of mobile or stationary SNs using mobile DS or static GW. WSN using LoRa and LDCA increases network capacity and coverage area reducing the required number of DS. It also reduces clustering energy to 33% and shows clustering efficiency of up to 98% for single-hop clustering covering 100 SNs. LoRa is not suitable for a clustered WSN with multiple hops due to its long TOA, depending on the LoRa link configurations (bandwidth and spreading factor). This research proposes a channel synchronization algorithm to improve the data rate of the LoRa link by combining multiple LoRa radio channels in a single logical channel. This increased data rate will enhance the capacity of the clusters in the WSN supporting faster clustering with mobile sensor nodes and data sink. Along with the LDCA, the proposed Lightweight Synchronization Algorithm for Quasi-orthogonal LoRa channels (LSAQ) facilitating multi-hop data transfer increases WSN capacity and coverage area. This research investigates quasi-orthogonality features of LoRa in terms of radio channel frequency, spreading factor (SF) and bandwidth. It derived mathematical models to obtain the optimal LoRa parameters for parallel data transmission using multiple SFs and developed a synchronization algorithm for LSAQ. The proposed LSAQ achieves up to a 46% improvement in network capacity and 58% in data rate compared with the WSN using the traditional LoRa Medium Access Control (MAC) layer protocols. Besides the high-density clustered WSN, remote monitoring applications like plant phenotyping may require transferring image or high-volume data using LoRa links. Wireless data transmission protocols used for high-volume data transmission using the link with a low data rate (like LoRa) requiring multiple packets create a significant amount of packet overload. Besides, the reliability of these data transmission protocols is highly dependent on acknowledgement (ACK) messages creating extra load on overall data transmission and hence reducing the application-specific effective data rate (goodput). This research proposes an application layer protocol to improve the goodput while transferring an image or sequential data over the LoRa links in the WSN. It uses dynamic acknowledgement (DACK) protocol for the LoRa physical layer to reduce the ACK message overhead. DACK uses end-of-transmission ACK messaging and transmits multiple packets as a block. It retransmits missing packets after receiving the ACK message at the end of multiple blocks. The goodput depends on the block size and the number of lossy packets that need to be retransmitted. It shows that the DACK LoRa can reduce the total ACK time 10 to 30 times comparing stop-wait protocol and ten times comparing multi-packet ACK protocol. The focused wide-area WSN and mobility requires different matrices to be evaluated. The performance evaluation matrices used for the static WSN do not consider the mobility and the related parameters, such as clustering efficiency in the network and hence cannot evaluate the performance of the proposed wide-area WSN platform supporting mobility. Therefore, new, and modified performance matrices are proposed to measure dynamic performance. It can measure the real-time clustering performance using the mobile data sink and sensor nodes, the cluster size, the coverage area of the WSN and more. All required hardware and software design, dimensioning, and performance evaluation models are also presented.

https://harvest.usask.ca/bitstreams/2c040e82-e73b-4f21-8a62-37c088457902/download