Implementing Efficient and Multi-Hop Image Acquisition In Remote Monitoring IoT systems using LoRa Technology

Abstract

Remote sensing or monitoring through the deployment of wireless sensor networks (WSNs) is considered an economical and convenient manner in which to collect information without cumbersome human intervention. Unfortunately, due to challenging deployment conditions, such as large geographic area, and lack of electricity and network infrastructure, designing such wireless sensor networks for large-scale farms or forests is difficult and expensive. Many WSN-appropriate wireless technologies, such as Wi-Fi, Bluetooth, Zigbee and 6LoWPAN, have been widely adopted in remote sensing. The performance of these technologies, however, is not sufficient for use across large areas. Generally, as the geographical scope expands, more devices need to be employed to expand network coverage, so the number and cost of devices in wireless sensor networks will increase dramatically. Besides, this type of deployment usually not only has a high probability of failure and high transmission costs, but also imposes additional overhead on system management and maintenance.

LoRa is an emerging physical layer standard for long range wireless communication. By utilizing chirp spread spectrum modulation, LoRa features a long communication range and broad signal coverage. At the same time, LoRa also has low power consumption. Thus, LoRa outperforms similar technologies in terms of hardware cost, power consumption and radio coverage. It is also considered to be one of the promising solutions for the future of the Internet of Things (IoT). As the research and development of LoRa are still in its early stages, it lacks sufficient support for multi-packet transport and complex deployment topologies. Therefore, LoRa is not able to further expand its network coverage and efficiently support big data transfers like other conventional technologies. Besides, due to the smaller payload and data rate in LoRa physical design, it is more challenging to implement these features in LoRa. These shortcomings limit the potential for LoRa to be used in more productive application scenarios.

This thesis addresses the problem of multi-packet and multi-hop transmission using LoRa by proposing two novel protocols, namely Multi-Packet LoRa (MPLR) and Multi-Hop LoRa (MHLR). LoRa's ability to transmit large messages is first evaluated in this thesis, and then the protocols are well designed and implemented to enrich LoRa's possibilities in image transmission applications and multi-hop topologies. MPLR introduces a reliable transport mechanism for multi-packet sensory data, making its network not limited to the transmission of small sensor data only. In collaboration with a data channel reservation technique, MPLR is able to greatly mitigate data collisions caused by the increased transmission time in laboratory experiments. MHLR realizes efficient routing in LoRa multi-hop transmission by utilizing the power of machine learning. The results of both indoor and outdoor experiments show that the machine learning based routing is effective in wireless sensor networks.