AI-Based Wireless Communication

AI-Based Wireless Communication

Sanjana T,, Lalitha S., Surendra H. H., Madhusudhan .. K. N.
Copyright: © 2022 |Pages: 19
DOI: 10.4018/978-1-6684-3804-6.ch004
(Individual Chapters)
No Current Special Offers


Artificial intelligence (AI) is one of the key enablers among quantum technology, smart meta-surfaces, dense antenna arrays, and mobile edge communication in 6G. The level of maturity achieved in the field of AI and development of computationally efficient hardware architectures with reduced costs have powered up the use of AI in different layers of wireless communication. Based on the learning, reasoning, and decision-making capability of AI, performance of wireless communication can be optimized. In addition, a whole new range of smart applications such as augmented reality (AR), virtual reality (VR), unmanned aerial vehicle (UAV), extended reality (XR) and holography, and autonomous driving, which demands high precision and low latency, can easily be accomplished by integrating AI into wireless communication. This chapter covers the role of AI in different layers, utilization of deep unfolding in physical layer, AI in mobile edge computing, explainable AI, federated learning, and AI for energy-efficient communication. The chapter concludes with research challenges and opportunities.
Chapter Preview


The advent of technologically powerful applications such as Autonomous driving systems in Automobile, Internet of Everything, Industrial IoT and connected robotics in industries, Holographic Communications, multi-sensory networking, gaming in Entertainment sector, Augmented reality, virtual reality in Entertainment and Education sector, Wireless brain and computer interactions, heterogeneous networks operating in smart environments, remote surgery in healthcare sector have been the driving forces for 6G communication.

The different features offered by 6G include data rate of 1Tbps, low latency upto 1ms, maximum spectral efficiency, high mobility support, and AI empowered. It supports machine to machine and machine to human interactions, Internet to Everything dense networks (Jagannath, 2021), smart environments and cell-free massive MIMO. In order to provide very high bandwidth, communications are pushed to THz and mm-wave frequencies. The key features of 6G include Computation Oriented Communications (Edge intelligence), Contextually Agile enhanced Mobile Broadband communications (situational awareness) and Event Defined ultra-reliable low latency communications (K. B. Letaief, 2019).

Edge intelligence will play a key role in Intelligent IoT as it adds trust, resilience, monitoring and detection, reliability and guarantees efficiency and value-added services in the network. Depending on the computing resources and storage at the edge devices there are different cases that could be realized. For example, training the model can be either done at the edge or cloud, and also its inference can be either done at the edge or device itself. This would definitely affect the quality of service provided to the end users. AI can be realized for the edge services or on the edge. AI for edge services includes managing energy efficiency and optimization based on location. AI on the edge includes data intelligence, computing, satisfying real-time requirements (Peltonen, 2020). The AI algorithms in 6G networks are generally applicable at the edge, or on different layers of wireless communication or in mobile applications or in self-configuring networks. The various network entities that would be needed to realize AI applications include caching, computing, wireless power transfer and communication. AI would assist in design and optimization of networks, network monitoring and management, radio resource management and security (Md Arifur Rahman, 2019). Artificial intelligence (AI) plays an important role as it not only contributes in design and optimization of core network, but also provides exciting services and applications for the end devices. Strategy Analytic have predicted that by 2023, 80% of the smart phones will be AI enabled (Strategy Analytics, 2021).

The highly dynamic nature of environment and huge volume and variety of data generated motivates the use of AI from network edge to the core. AI makes the 6G architecture intelligent in terms of flexibility, adaptivity, computational, fast learning, reasoning, smart decision making and agile. It is called Intelligent Radio (IR) as the operating system which exists between hardware and transceiver algorithms, is capable of estimating the abilities of hardware and based on which configures the algorithms (K. B. Letaief, 2019).

Key Terms in this Chapter

Physical Layer: It is most important to have great capabilities at the lower layers in order to support the upper layers to have an overall efficient communication system. Physical layer mostly involves signal processing tasks suitable for effective transmission.

Distributed AI: It simplifies complex tasks by utilizing spatial distribution of computing resources. It is very much helpful to process large data sets using distributed nodes.

Deep Reinforcement Learning: It is a combination of DL and RL. DRL is effective in several applications.

AI in Mobile Edge Computing: It is a unique feature of 6G where most of the processing is moved from cloud to edge devices because of their enhanced capabilities.

Federated Learning: It is decentralized approach employed at the edge nodes which learns global model collaboratively using local data sets. It is basically a distributed training technique.

Explainable AI: It is essential to develop trust in the system when decisions are taken by AI techniques. The techniques should be transparent in terms of operations performed and reasoning for the decisions taken.

Challenges: Several key enablers of 6G such as AI, quantum technology, new frequency bands, smart meta-surfaces, dense antenna arrays and mobile edge communication have benefits as well challenges associated. Challenges in terms of design and implementation leads to new research opportunities.

Deep Unfolding: It is a technique of using Neural Networks to implement complex iterative signal processing approaches in a simplified way especially at the physical layer.

Complete Chapter List

Search this Book: