Search

Performance monitoring and state estimation are promising techniques to ensure high reliability in Internet of Things (IoT). A major data-driven approach that addresses the decentralized and heterogeneous nature of IoT is distributed computing. Values from physical measurements and temporal dimension in transmitted sensor data can be analyzed in a distributed manner for state estimation. In this paper, we discuss a smart home energy utilization problem and the effects of climatic conditions. We focus on determining patterns, correlations and predicting weather attributes in a distributed and scalable manner using Message Passing Interface (MPI). Our results show that our distributed approach is scalable up to 24 nodes or processes, which can further scale up to 1000 nodes or processes depending on the number of indivisible subtasks that need to be computed. We also compare the execution time for the distributed tasks and notice a performance gain due to parallelization. Furthermore, we develop a model based on Long Short-Term Memory (LSTM) to estimate the expected energy to be utilized in a smart home environment. We evaluate our work using an IoT dataset to show the effectiveness of our approach.

With the growth of Internet of Things (IoT) and mobile edge computing, billions of smart devices are interconnected to develop applications used in various domains including smart homes, healthcare and smart manufacturing. Deep learning has been extensively utilized in various IoT applications which require huge amount of data for model training. Due to privacy requirements, smart IoT devices do not release data to a remote third party for their use. To overcome this problem, collaborative approach to deep learning, also known as Collaborative Deep Learning (CDL) has been largely employed in data-driven applications. This approach enables multiple edge IoT devices to train their models locally on mobile edge devices. In this paper, we address IoT device training problem in CDL by analyzing the behavior of mobile edge devices using a game-theoretic model, where each mobile edge device aims at maximizing the accuracy of its local model at the same time limiting the overhead of participating in CDL. We analyze the Nash Equilibrium in an N-player static game model. We further present a novel clusterbased fair strategy to approximately solve the CDL game to enforce mobile edge devices for cooperation. Our experimental results and evaluation analysis in a real-world smart home deployment show that 80% mobile edge devices are ready to cooperate in CDL, while 20% of them do not train their local models collaboratively.

Data privacy and security of Internet enabled devices has become a major concern of many users and manufacturers. The proliferation of Internet of Things (IoT) devices and the increasing network traffic have heightened the attack surface. Proxy and man-in-the-middle attacks can be used to exploit vulnerability inherent in IoT devices. Due to the limited resources and features of IoT devices, transmitted data might be at risk of traffic interception and possible decryption of encrypted data on the fly using these attacks. In this paper, we discuss a vulnerability in IoT communication and propose an effective approach based on Deep Q-Network (DQN) and Generative Adversarial Network (GAN) for proxy detection. Our main aim is a robust detection mechanism using network connection information. We further propose a use case for distributed machine learning suitable for real-time monitoring and proxy detection.