No abstract

No abstract

Advancement in IoT technology and the concept of Information-Centric Networking lead to less importance of node individuality since several nodes can work interchangeably. Multiple sensor nodes can be grouped into a region and monitored as one instance to guarantee sufficient coverage over the region. Therefore, a single node fault often does not need to be reported unless it is the last node in the region. In addition, there are occasions where a central monitor station cannot rely on continuous data delivery from nodes or regions to decide whether they are still alive, such as situations when nodes are deployed to detect rare events. Moreover, low-power listening MAC protocols, which significantly help reducing power consumption while nodes are mostly idle, put a lot more work on the transmission process. In such situations it is desirable to minimize status reports to the central monitor station. A distributed region-based monitoring scheme, or DRMON, is then proposed to facilitate this circumstance. This approach designates a representative to each region so that it can be used as an indicator of the region's status with a mechanism to re-elect a new representative until all nodes in the respective region are dead, implying region inactiveness. We evaluate the suitability of DRMON over various scenarios in two aspects: centralized vs. distributed monitoring schemes and individual-based vs. region-based monitoring schemes. Simulation results indicate that region-based schemes outperform the individual schemes in terms of power consumption and scalability when the number of regions is low. The distributed schemes also yield better efficiency in terms of message overhead and load distribution. In addition, detection accuracy of all schemes is not significantly different and fault detection delay is guaranteed. This outcome suggests that in the case where existence of individual node is out of concern, distributed region-based fault monitoring scheme could be employed to reduce energy usage and lower message overhead while retaining the detection accuracy.

Advancement in IoT technology and the concept of Information-Centric Networking lead to less importance of node individuality since several nodes can work interchangeably. Multiple sensor nodes can be grouped into a region and monitored as one instance to guarantee su°cient coverage over the region. Therefore, a single node fault often does not need to be reported unless it is the last node in the region. In addition, applications focusing on detecting rare events rarely require nodes to transmit and often rely on a low-power listening MAC protocols, where nodes spend most of their time sleeping but require signi˝cantly more work during transmission. In such situations it is desirable to avoid periodic status reports transmitted to the central monitor station as usually found in a centralized monitoring scheme. A distributed region-based monitoring scheme, or DRMON, is then proposed to facilitate this circumstance. This approach designates a representative to each region so that it can be used as an indicator of the region's status with a mechanism to re-elect a new representative until all nodes in the respective region are dead, implying region inactiveness. We evaluate the suitability of DRMON over various scenarios in two aspects: centralized vs. distributed monitoring schemes and individual-based vs. region-based monitoring schemes, along with existing work in the literature. Simulation results indicate that region-based schemes outperform the individual schemes in terms of power consumption and scalability when the number of regions is low. The distributed schemes also yield better e°-ciency in terms of message overhead. Compared against the other schemes, DRMON's overall power consumption is reduced by 4%-10%, with 66%- 88% reduction in packet transmissions, while maintaining fault detection precision and recall of greater than 90% and the detection delay within an acceptable range. This outcome suggests that in the case where existence of individual node is out of concern, distributed region-based fault monitoring scheme could be employed to reduce energy usage and lower message overhead while retaining acceptable detection accuracy.