Artificial intelligence (AI) and machine learning (ML) are widely employed to make the solutions more accurate and autonomous in many smart and intelligent applications in the Internet of Things (IoT). In these IoT applications, the performance and accuracy of AI/ML models are the main concerns; however, the transparency, interpretability, and responsibility of the models’ decisions are often neglected. Moreover, in AI/ML-supported next-generation IoT applications, there is a need for more reliable, transparent, and explainable systems. In particular, regardless of whether the decisions are simple or complex, how the decision is made, which features affect the decision, and their adoption and interpretation by people or experts are crucial issues. Also, people typically perceive unpredictable or opaque AI outcomes with skepticism, which reduces the adoption and proliferation of IoT applications. To that end, explainable AI (XAI) has emerged as a promising research topic that allows ante-hoc and post-hoc functioning and stages of black-box models to be transparent, understandable, and interpretable. In this article, we provide an in-depth and systematic review of recent studies that use XAI models in the scope of the IoT domain. We classify the studies according to their methodology and application areas. Additionally, we highlight the challenges and open issues and provide promising future directions to lead the researchers in future investigations.

This paper proposes a novel quantum computing-inspired approach to multi-objective optimization, called Quantum Computing Inspired Non-dominated Sorting Genetic Algorithm II (QNSGA-II). Although Non-dominated Sorting Genetic Algorithm II (NSGA-II) has been effectively used in the literature to solve a variety of optimization issues, it may encounter some difficulties especially in handling heavily constrained problems due to its premature convergence. The proposed approach mitigates this difficulty by combining conventional NSGA-II with the concept and principles of quantum computing. QNSGA-II exploits quantum bits and superposition of states to reduce the convergence time and improve search space capability by evolving the probabilistic model. This paper aims to provide more detailed information about our proposed algorithm and its advantages.

With the evolution of artificial intelligence, Machine Learning (ML) techniques have become more powerful predictors, and accordingly, the use of ML techniques has become a part of our daily life in different application scenarios such as disease diagnosis, movie recommendation, monitoring system, or detection of malicious attacks. Although ML provides high accurate predictions, it suffers from opacity. By behaving like a black box it excluded users about how to reach particular decisions. Interpretable Machine Learning (IML) is a recent technology that offers a promising solution to the opaqueness problem of complex ML techniques. It provides transparency of how the inner workings of ML lead to certain decisions and allows users to be aware of the decision-making process. Especially, in critical scenarios such as healthcare, it may become extremely important to know the reasons that affect the decision as well as the result. In this study, we aim to show the benefits of IML over a healthcare case study. In experiments, we employ SHAP and LIME IML models for the Random Forest (RF) and Gradient Boosting (GB) algorithms for the problem of diagnosing diabetes and its explanations. Overall results exhibit that applying IML models to complex and hard-to-interpret ML techniques ensures detailed interpretability while maintaining accuracy. We also perform experiments for local interpretability by focusing on an instance, which is another advantage of IML.

Real-time tracking of a maneuvering aircraft is a challenging issue in the literature. For effective tracking, an accurate and complete transfer of data from aviation to ground systems is needed. However, possible data loss caused by telemetry or data acquisition system makes aircraft tracking difficult. To mitigate this problem, efficient tracking systems are proposed in the literature. Accordingly, we first present a brief survey of aircraft tracking systems by splitting them according to their approaches as mathematical, machine learning-based, and deep learning-based approaches. After examining the existing studies, we offer a real-time Deep learning-based Aircraft Tracking (DeepAT) system that enables real-time tracking of an aircraft. Deep learning models are employed to predict the next location of aircraft. Accordingly, the radar angle of the antenna is determined to the antenna or radar system directs to the next location. When the proposed model is analyzed through two potential use case scenarios which are flight tests and military applications, DeepAT offers promising solutions to prevent data loss in different applications.

Recent online popular platforms such as social media, blogs, and newspapers generate a vast amount of unstructured data per second. Sentiment Analysis (SA) is an efficient technique to identify and extract subjective information in unstructured data to enable businesses to understand the emotional tendency of the interactive users towards its products or services. However, analyzing unstructured data can be more difficult than structural data. In particular, the performance of SA techniques decreases due to the structural complexity of the language. SA techniques are widely used in English since it is universal and structurally more suitable for SA. On the other hand, structural difficulties and complexities in Turkish cause performance degradation of SA studies compared to English. This study aims to overcome this difficulty by first translating Turkish texts into English texts by machine translation, and then realizing sentiment analysis on English texts. To demonstrate the success of machine translation, the experiments are conducted on two different data sets and results are given in a comparative manner for both on Turkish as the original language and English as the translated language. Data sets in both languages are classified by six different machine learning methods which are Logistic Regression, Naive Bayes, Decision Tree, Random Forest, Support Vector Machine, and Artificial Neural Network. When the success rates of machine learning methods are examined, a significant increase is observed by machine translation for most of the methods.

With the proliferation of machine-to-machine communication, there are many communication protocols standardized for IoT applications. Performances of these protocols may significantly deviate from each other even under the same operating conditions. In this paper, we quantitatively compare the performances of a set of well-known IoT communication protocols, namely CoAP (Constrained Application Protocol), MQTT (Message Queuing Telemetry Transport) and XMPP (Extendible Message Persistent Protocol) in a real-world testbed. CoAP employs UDP packets for transmission while others use TCP. For this purpose, we design as small testbed that collects real-time environmental data. By designing such a system, we aim to reveal the differences among protocols in terms of packet creation time and packet transmission time. The obtained results show that XMPP is worse than other protocols in both metrics and MQTT and CoAP perform almost equally.

Fog computing is a promising technology that helps overcome to the difficulties of handling a huge amount of Internet of Things (IoT) data by distributing its applications and services closer to the network edge. In fog computing, distributed fog servers’ proximity to the end devices enables upcoming and outgoing data to be forwarded efficiently. However, as the size of fog-enabled IoT platforms increases, efficient routing mechanisms are required in fog computing to forward the ubiquitous IoT data to related servers with low latency, low bandwidth, and/or high security. Therefore, software defined networking (SDN) can be employed to optimize the routing process in fog-enabled IoT platforms. In this paper, we first identify the routing requirements of fog computing. Then, we present an extensive survey of routing in fog-enabled IoT platforms. Lastly, we propose a hierarchical SDN-based fog computing architecture for routing in fog-enabled IoT platforms where fog controllers take actions locally for frequent events while the cloud controller takes actions globally for rare events. The proposed framework is evaluated by varying the number of controllers. The initial results show that the proposed hierarchical SDN-based framework has the potential to reduce routing delay and data transmission overhead. Also, throughput is increased as the number of controllers increases.

Traditional electric generation based on fossil fuel consumption threatens the humanity with global warming, climate change, and increased carbon emission. Renewable resources such as wind or solar power are the solution to these problems. The smart grid is the only choice to integrate green power resources into the energy distribution system, control power usage, and balance energy load. Smart grids employ smart meters which are responsible for two-way flows of electricity information to monitor and manage the electricity consumption. In a large smart grid, smart meters produce tremendous amount of data that are hard to process, analyze and store even with cloud computing. Fog computing is an environment that offers a place for collecting, computing and storing smart meter data before transmitting them to the cloud. This environment acts as a bridge in the middle of the smart grid and the cloud. It is geographically distributed and overhauls cloud computing via additional capabilities including reduced latency, increased privacy and locality for smart grids. This study overviews fog computing in smart grids by analyzing its capabilities and issues. It presents the state-of-the-art in area, defines a fog computing based smart grid and, gives a use case scenario for the proposed model.