The Intersection Of ML and IoT
The Internet of Things (IoT) and the Industrial Internet of Things (IIoT) is evolving towards the next generation of Tactile IoT/IIoT, which will bring together hyperconnectivity, edge computing, Distributed Ledger Technologies (DLTs) and Artificial Intelligence (AI). Future IoT applications will apply AI methods, such as machine learning (ML) and neural networks (NNs), to optimize the processing of information, as well as to integrate robotic devices, drones, autonomous vehicles, augmented and virtual reality (AR/VR), and digital assistants. These applications will engender new products, services and experiences that will offer many benefits to businesses, consumers and industries. A more human-centered perspective will allow us to maximize the effects of the next generation of IoT/IIoT technologies and applications as we move towards the integration of intelligent objects with social capabilities that need to address the interactions between autonomous systems and humans in a seamless way.
Next Generation IoT technology convergence
The IoT is enabled by heterogeneous technologies used to sense, collect, store, act, process, infer, transmit, create notifications of/for, manage and analyse data. The combination of emergent technologies for information processing and distributed security, e.g. AI, IoT, DLTs and blockchains, brings new challenges in addressing distributed IoT architectures and distributed security mechanisms that form the foundation of improved and, eventually, entirely new products and services. New systems in the IoT that use smart solutions with embedded intelligence, connectivity and processing capabilities for edge devices rely on real-time analysis of information at the edge. These new IoT systems are moving away from centralized cloud-computing solutions towards distributed intelligent edge computing systems. Traditional centralized cloud computing solutions are perfect for non-real-time applications that require high data rates, huge amounts of storage and processing power, are not strict to very low latency, cost money and can be used for heavy data analytics and AI processing jobs. On the other hand, distributed edge solutions introduce computations at the edge of the network where information is generated and are perfect for real-time services, since they exhibit very low latency (in the order of milliseconds) and can be used for simple ultra-fast analytics jobs. The collection, storage and processing of data at the edge of the network in a distributed way contributes also to the increased privacy of the user data, since no personal information is stored in backbone centralized servers and each user retains the full control of his data.
IoT Strategic Research and Innovation
The Internet of Things European Research Cluster (IERC) concentrates the know-how regarding scientific production and research capacity for the Internet of Things in Europe; the IERC brings together EU-funded projects with the aim of defining a common vision for IoT technology and addressing European research challenges. The rationale is to leverage the large potential for IoT-based capabilities and promote the use of the results of existing projects to encourage the convergence of ongoing work; ultimately, the endpoints are to tackle the most important deployment issues, transfer research and knowledge to products and services, and apply these to real IoT applications. The objectives of IERC are to provide information on research and innovation trends, and to present the state of the art in terms of IoT technology and societal analysis, to apply developments to IoT-funded projects and to market applications and EU policies. The final goal is to test and develop innovative and interoperable IoT solutions in areas of industrial and public interest.
List of the main open research challenges for the future of IoT:
• IoT architectures considering the requirements of distributed intelligence at the edge, cognition, artificial intelligence, context awareness, tactile applications, heterogeneous devices, end-to-end security, privacy, trust, safety and reliability.
• IoT systems architectures integrated with network architecture forming a knowledge-centric network for IoT.
• Intelligence and context awareness at the IoT edge, using advanced distributed predictive analytics.
• IoT applications that anticipate human and machine behaviours for social support.
• Tactile Internet of Things applications and supportive technologies.
• Augmented reality and virtual reality IoT applications.
• Autonomics in IoT towards the Internet of Autonomous Things.
• Inclusion of robotics in the IoT towards the Internet of Robotic Things. 30 The Next Generation Internet of Things – Hyperconnectivity
• Artificial intelligence and machine learning mechanisms for automating IoT processes.
• Distributed IoT systems using securely interconnected and synchronized mobile edge IoT clouds.
• Stronger distributed and end-to-end holistic security solutions for IoT, preventing the exploitation of IoT devices for launching cyber-attacks, i.e., remotely controlling IoT devices for launching Distributed Denial of Service (DDoS) attacks.
• Stronger privacy solutions, considering the requirements of the new General Data Protection Regulation (GDPR)  for protecting the users’ personal data from unauthorized access, employing protective measures (such as Privacy Enhancing Technologies – PETs) as closer to the user as possible.
• Cross-layer optimization of networking, analytics, security, communication and intelligence.
• IoT-specific heterogeneous networking technologies that consider the diverse requirements of IoT applications, mobile IoT devices, delay tolerant networks, energy consumption, bidirectional communication interfaces that dynamically change characteristics to adapt to application needs, dynamic spectrum access for wireless devices, and multi-radio IoT devices.
• Adaptation of software defined radio and software defined networking technologies in the IoT.
The Tactile IoT/IIoT is a shift in the collaborative paradigm, adding humancentred perspective and sensing/actuating capabilities transported over the network to communications modalities, so that people and machines no longer need to be physically close to the systems they operate or interact with as they can be controlled remotely. Tactile IoT/IIoT combines ultra-low latency with extremely high availability, reliability and security and enables humans and machines to interact with their environment, in real-time, using haptic interaction with visual feedback, while on the move and within a certain spatial communication range.
The research challenges for implementing AI at the edge of networks for IoT applications areas follows:
• Mechanisms for collecting and aggregating data and information and developing edge models that generate insights from the data available in real-time by providing methods and techniques to train models in the edge environment with appropriately distributed storage capabilities
• ‘AI-friendly’ processors to address the AI workloads for IoT applications requiring AI computationally-intensive capabilities; research and development concerning architectural concepts to shift central control to the edge and the use of modified graphics processor units, hybrid processors and AI-based processors, embedding accelerators and neural networks for processing specific AI algorithms
• New energy- and resource-efficient methods for image recognition and geospatial processing using AI at the edge, based on machine learning and other AI techniques
• Edge computing implementation based on neuromorphic computing and in-memory computing to process unstructured data, such as images or video, used in IoT applications.
• Edge computing implementations based on distributed approaches for IoT computing systems at the edge
• Distributed IoT end-to-end security for AI-based solutions that process data at the edge using a group of edge nodes to work together on a particular task, thereby ensuring that no security holes or attacks are possible
• AI for smart data storage in edge-based IoT
• AI for software-defined networking in edge-based IoT
• Swarm intelligence algorithms for edge-based IoT/IIoT • Machine learning, deep learning and multi-agent systems for edge-based IoT/IIoT
• Cognitive aspects of AI in edge-based IoT/IIoT
• Neural networks for AI in edge-based IoT/IIoT
• Distributed heterogeneous memory systems design for AI in edge-based IoT/IIoT
Networks and Communication
It is predicted that the adoption of low-power short-range networks for wireless IoT connectivity will increase through 2025 and will coexist with wide-area IoT networks], while 5G networks will deliver 1,000 to 5,000 times more capacity than 3G and 4G networks today. IoT technologies are extending known business models, leading to the proliferation of different ones as companies push beyond the data, analytics and intelligence boundaries. IoT devices will be contributing to and strongly driving this development.
Changes will first be embedded in given communication standards and networks and subsequently in the communication and network structures defined by these standards. 5G and the IoT promise new capabilities and use cases, which are set to impact not only consumer services but also many industries embarking on their digital transformations. New massive IoT cellular technologies, such as NB-IoT and Cat-M1, are taking off and driving growth in the number of cellular IoT connections, with a CAGR of 30 percent expected between 2017 and 2023. These complementary technologies support diverse LPWAN use cases over the same underlying LTE network.
Please like, Share and Comments