The Internet of Things (IoT) environments have hard realtime tasks that need execution within fixed deadlines. As IoT devices consist of a myriad of sensors, each task is composed of multiple interdependent subtasks. Towards this, the cloud and fog computing platforms have the potential of facilitating these IoT Sensor Nodes (SNs) in accommodating complex operations with minimum delay. To further reduce operational latencies, we breakdown the high-level tasks into smaller subtasks and form a Directed Acyclic Task Graph (DATG). Initially, the SNs offload their tasks to a nearby Fog Node (FN) based on a greedy choice. The greedy formulation helps in selecting the FN in linear time while avoiding combinatorial optimizations at the SN, which saves time as well as energy. IoT environments are highly dynamic, which mandates the need for adaptive solutions. At the chosen FN, depending on the dependencies on the DATGs, its corresponding deadlines, and the varying conditions of the other FNs, we propose an epsilon-greedy nonstationary multiarmed bandit-based scheme (D2CIT) for online task allocation among them. The online learning D2CIT scheme allows the FN to autonomously select a set of FNs for distributing the subtasks among themselves and executes the subtasks in parallel with minimum latency, energy, and resource usage. Simulation results show that D2CIT offers a reduction in latency by 17% compared to traditional fog computing schemes. Additionally, upon comparison with existing online learningbased task offloading solutions in fog environments, D2CIT offers an improved speedup of 59% due to the induced parallelism.

Integration of various technologies to an Internet of Things (IoT) framework share the common goals of a consistent and structured data format that can be applied to any device, given the vast application scope of IoT. Additional goals include minimizing channel traffic and system energy consumption. In this work, we propose to dismiss the requirement of certain seemingly crucial identifier fields from packets arriving through various sensor nodes in an agricultural IoT deployment. The proposed approach reduces packet size, thereby reducing channel traffic and energy consumption, as well as retaining the capability of identifying these originating nodes. We propose a method of a blind agricultural IoT node and sensor identification, which can be sourced and operated from a master node as well as a remote server. Additionally, this scheme has the capability of detecting the radio link quality between the master and slave nodes in a rudimentary form, as well as identifying the sensor nodes. We successfully trained and tested various multi-layer perceptron (MLP)-based models for blind identification, in real-time, using our implemented agricultural IoT implementation. The effect of changes in learning rate and momentum of the optimizer on the accuracy of classification is also studied. The projected cumulative energy savings across the network architecture, of our scheme, in conjunction with TCP/IP header compression techniques, are substantial. For a 100 node deployment using a combination of the proposed blind identification reduced sampling strategies over regular IPv4-based TCP/IP connection, an estimated annual saving of 99% is projected.

Dense IoT implementations incur heavy data load on the implemented networks. In this paper, we propose and evaluate a low-latency method of increasing the packet throughput in agricultural IoT implementations. The proposed method envisions removal of node identifiers from packets before transmission and predictive packet-source mapping method within the edge layer of an agrarian Internet of Things (IoT) implementation. The edge layer following a master-slave architecture. Pre-trained lightweight machine learning models at the edge identify the origin of the incoming packets based on the long-term learned collective variations of the sensorial values from the slave node. This reduction in packets significantly frees up time-slots at the receiving master node, allowing for more simultaneous connections to it. This intra-edge packet origin mapping scheme is further compared with the approach of edge node identification at a remote server to adjudge the tradeoffs between accuracy and latency of transmission. The proposed method doubles the amount of sensor data transmitted between the slave to master nodes with significant energy savings over longer duration and increases the data throughput by approximately 1:5 times between the master node and the remote server for our implementation. The proposed method estimates energy savings in the order of 20 watts for a deployment setup of 100 nodes over a year. The energy savings over densely deployed IoT networks can be utilized to accommodate more nodes and increase the lifetime of the network.

The utility of knowledge discovery on the Internet of Things (IoT) and its allied domains is undeniably one of the most indispensable ones, which results in optimized placement architectures, efficient routing protocols, device energy savings, and enhanced security measures for the implementation. The absence of knowledge discovery in IoT results in just an implementation of large-scale sensor networks, which generates a huge amount of data, and which needs, an often under-optimized, processing for actionable outputs. In this survey, we explore the various domains of IoT for which knowledge discovery is inseparable from the application, and how it benefits the overall implementation of the IoT architecture.

In this paper we propose SmartARM – a Smartphone-based group Activity Recognition and Monitoring (ARM) scheme, which is capable of recognizing and centrally monitoring coordinated group and individual group member activities of soldiers in the context of military excercises. In this implementation, we specifically consider military operations, where the group members perform similar motions or manoeuvres on a mission. Additionally, remote administrators at the command center receive data from the smartphones on a central server, enabling them to visualize and monitor the overall status of soldiers in situations such as battlefields, urban operations and during soldier’s physical training. This work establishes – (a) the optimum position of smartphone placement on a soldier, (b) the optimum classifier to use from a given set of options, and (c) the minimum sensors or sensor combinations to use for reliable detection of physical activities, while reducing the data-load on the network. The activity recognition modules using the selected classifiers are trained on available data-sets using a testtrain- validation split approach. The trained models are used for recognizing activities from live smartphone data. The proposed activity detection method puts forth an accuracy of 80% for real-time data.

The use of rovers in navigating dangerous and uncharted terrains for localization, mapping and data-gathering for search and rescue in disaster management is a challenging domain, especially with constrained network connectivity. In this paper, a rover mounted rotating turret-based 2D ultrasonic distance mapping and Inertial Measurement Unit (IMU)-based dead-reckoning (R2D2) mechanism is proposed. The wireless rover is operated over a constrained network, allowing its handler to gather situational awareness of a hazardous site remotely, and without risking lives. A visualization of the path of this rover by means of a low data intensity fusion map generated using a combination of ultrasonic distance measurements and relative positioning information from the dead-reckoning system is performed by the remote operator