Requirements for Next Generation Distributed Intelligence Wireless Networks

The move toward distributed intelligent networks using 5G and AIoT systems would force the entire telecommunications industry to evolve to cater to the growing demand across all verticals. The convergence of 5G and AI will allow the development of distributed intelligent infrastructure that permeates autonomous decisionmaking process across various industries and applications in real time. Such flexible and adaptable intelligent networks w'ould enable:

  • 1. Scalability: Using network slicing and orchestration techniques, 5G intelligent networks w'ould be able to cater to different service types and applications with different quality of service requirements. For example, network slice dedicated to healthcare services can be given utmost priority among all the use cases.
  • 2. Decentralized Intelligence: By shifting AI compute capabilities away from central nodes toward the edge, intelligent macro and micro cells will bring about a paradigm shift of cloud computing closer to the end user. Intelligent 5G networks would be expected to operate in such highly dense deployments, thereby improving performance by enabling intelligence to be distributed across the entire network from the cloud to the IoT edge. This would have the added benefit of improving service experiences and would enhance the overall infrastructure efficiency.
  • 3. Enhanced Operational Efficiency: Utilizing Al-assisted intelligent 5G networks w'ould allow for several infrastructure processes to be automated, thus, enabling industries to minimize waste, make smart decisions and thereby increase production and operational efficiency by reducing human intervention and simplifying tech complexities.
  • 4. Improved Network Security: Utilizing distributed intelligence in 5G networks will allow analysis of a large quantity of data, thereby enabling much effective detection and defense against malicious network attacks. For example, AI
On-device contextual and environmental sensing to reduce network access overhead and latency

FIGURE 8.3 On-device contextual and environmental sensing to reduce network access overhead and latency.

could detect anomalies in network traffic (e.g., flooding, jamming) by analyzing unusual spectrum usage patterns and automatically take corrective steps without human intervention.

5. Radio Awareness: On-device AI inference would allow AIoT devices contextual and environmental sensing in complex RF environments that would improve 5G end-to-end systems by reducing network data traffic for efficient wireless mobility, spectrum utilization and improved radio security, shown in Figure 8.3 (Smee and Hou. 2020). Such devices would be capable of intelligent beamforming and power management, thereby improving data throughput and increase battery life and robustness.

Enabling Technological Uses Cases for 5G and AIoT Systems

Distributed intelligent wireless networks have the potential to become the technological platform for a new wave of innovations and use cases. Utilizing ultra-low- latency and high-bandwidth 5G spectrum and on-device AI inference, wireless AIoT edge devices will enable flexible solutions for different verticals and use cases. Such an architecture would be able to adapt to different network environments and allow for appropriate economic and performance trade-offs to achieve the optimum requirements for different applications. The convergence of 5G and AIoT is expected to play a major role in five key domains, namely industrial manufacturing, transportation and logistics, healthcare, security and the entertainment industry. Below we discuss a few use cases in each domain based on how the convergence of 5G and AIoT can enhance productivity.

8.4.1 Industry 4.0

In the industrial manufacturing segment, distributed intelligent 5G networks will be instrumental for implementing the next industrial revolution coined as Industry 4.0 or even as Industrial Internet of Things (IIoT) (Wollschlaeger, Sauter, and Jasperneite, 2017; Aijaz and Sooriyabandara 2018). Enabling optimized logistics, automated factories, and remotely operated machinery would significantly trim production downtime, allow for automated real-time preventive maintenance, and enhance quality assurance, thereby increasing productivity.

Use Case 1: Automated Factories and Remote Inspection

Amalgamation of ultra-fast and ultra-low-latency 5G connectivity with intelligent AIoT devices will enable enhanced automation of several industrial processes and machinery (Wollschlaeger et al„ 2017). For example, connected sensors and cameras with onboard AI could automatically predict manufacturing defects in a production line and take corrective decisions in real time. 5G connectivity would allow seamless remote supervision and control of industrial robots with real-time visual and haptic feedback using interactive tools such as smart gloves and smart headsets with virtual/ augmented reality (Aijaz and Sooriyabandara, 2018). This would result in lowering costs and minimizing operational risks in hazardous areas such as nuclear power plants, off-shore oil rigs, etc.

8.4.2 Transportation and Logistics

In the automotive industry, AI is already transforming vehicular automation and transportation and logistics management. Using onboard sensors and cameras in conjunction with Al-assisted machine vision models, autonomous vehicles are now capable of self-driving, navigation, and even traffic monitoring in complex and dynamically altering environments (Zhao et ah, 2018; Tanwar et ah 2019). Using 5G in conjunction with AIoT would allow these applications to reach their full potential.

Use Case 1: Connected Intelligent Vehicles

Connected vehicles with onboard AIoT systems will be capable of self-navigation by contextually communicating wirelessly with each other and roadside systems using vehicle-to-everything (V2X) communication protocol over 5G networks in real time (Zhao et ah, 2018). This would also enable automated traffic management and eventually lead to the development of driverless public transport models that will improve public safety by minimizing traffic accidents. Fleets of connected and Unmanned Aerial Vehicles (UAVs) can also be used for logistics operations (Tanwar et ah, 2019; Ullah et ah, 2019). UAVs are currently being adopted for delivering goods in challenging terrains and even urban congestions. Using 5G networks and onboard AIoT systems would allow swarms of UAVs to self-coordinate, avoid collisions with each other, and avoid other obstacles along their path. Such systems would provide significant cost savings to government authorities and the end user as they w'ould be much cheaper to maintain than manned delivery systems.

8.4.3 HealthCare 5.0

Presently, AI in healthcare is deployed in diagnosis, drug synthesis, and patient screening. The next medical revolution, coined as Healthcare 5.0. will utilize super- low latency and high bandwidths and data rates of 5G networks in conjunction with

AIoT medical devices, for innovative applications such as real-time personalized patient monitoring and care (also known as precision medicine) and even aid medical practitioners to perform remote Al-assisted surgeries to name a few (Soldani et al., 2017; Mohanta, Das, and Patnaik, 2019; Ullah et al., 2019).

Use Case 1: Precision Medicine

The current generation of medical-grade edge devices such as electrocardiograms (ECGs) and glucose monitoring devices utilize trained Al algorithms to detect anomalies in patients’ vital signs for early diagnosis of any potential medical condition (Ullah et al., 2019). Utilizing 5G connectivity, it is possible for these Al-enabled devices to provide more effective care by facilitating remote monitoring and diagnosis. This technology has the potential to revolutionize medical care, which in some instances is limited to geographical location of medical practitioners. Adoption of wearable semi- or noninvasive biometric devices such as fitness trackers and on-skin sensors have gained popularity (Mohanta, Das, and Patnaik, 2019). The current generation of these medical devices can only perform data measurement and transmission to the user’s smartphone for data processing and notification. Future versions with integrated 5G radio and trained Al models would capable of continuously monitoring patients’ vital signs for detecting any medical anomalies and alert the concerned authorities for immediate medical care in real time.

Use Case 2: Remote Diagnosis and Surgeries

There are instances w'hen patients are either too far away from medical institutions or are in too critical a condition to be moved. The tangible nature of distributed intelligent networks enabled using ultra-reliable, ultra-high-speed and ultra-low-latency 5G networks would enable doctors to remotely diagnose such cases using audio video connectivity with real-time haptic feedback. Using 5G and AIoT-based robotic systems, it would also be possible for medical specialists to perform remote surgeries in the near future (Soldani et al., 2017).

8.4.4 Security and Safety

Distributed intelligent 5G networks have the potential to make our localities and cities much safer by allowing government institutions to fight crime. This can be achieved by improving security surveillance systems and emergency services while minimizing incurred cost (Horn and Schneider, 2015; Ahmad et al., 2018).

Use Case 1: Intelligent Surveillance

Distributed intelligent 5G networks will facilitate deployment of Al-integrated connected surveillance systems such as security alarms, cameras, and sensors. This would allow for real-time surveillance and automated assessment (Horn and Schneider, 2015). For example, AIoT-enabled security cameras will be capable of automatically analyzing human behavior (e.g., body language, facial expressions) to detect suspects in real time and alert the relevant authorities. In the future such systems would be able to predict offences w'ell in advance and thereby aid in optimizing the use of crime-prevention systems.

Use Case 2: Border/Immigration Control and Emergency Services

Swarms of autonomous or remotely operated drones with onboard AI inference cameras connected over 5G networks can aid deployment of controlled emergency services and disaster management and can also be used for border control (Ahmad et ah, 2018). Using ultra-fast 5G networks would aid emergency service personnel in operations in hazardous environments such as sites of toxic contamination and forest fires. They can also be used at border control sites to detect unauthorized trespassers and prevent crime.

8.4.5 Entertainment and Retail

The amalgamation of 5G networks and AIoT devices introduces several advantages for the media and entertainment industry. Innovative applications such as immersive augmented and/or virtual reality (Schmoll et ah, 2018) and 8K resolution content (Inoue et ah, 2017) have very high computational requirements. This inherently points toward the need for high-bandwidth 5G networks combined with AIoT edge devices to provide consumers with personalized entertainment experiences.

Use Case 1: 5G Cloud Gaming

Current online gaming experiences have been poor due to limited bandwidth and delays associated with our current 4G/LTE networks. Online gaming is an area where even a millisecond delay affects playability. It is predicted that intelligent 5G networks will play a revolutionary role in meeting these requirements (Braun et ah, 2017). Companies like Google (with Stadia) and Microsoft (with X-cloud) are rolling out their online gaming platforms based on the backbone of intelligent 5G networks. Utilizing ultra-low latency data transmission technologies, 5G networks will be able to reduce transmission delays down to less than 1 millisecond. Service providers could dedicate network slices dedicated for online gaming, thereby guaranteeing reliable service to gamers as per their contractual agreement with service providers. Distributed intelligence in 5G networks would then automatically allocate network resources to guarantee reliable performance to consumers in different peak times and days.

Use Case 2: Personalized Shopping Experience

Targeted advertising can be a highly effective form of advertising by providing every consumer with a tailored shopping experience, thereby increasing effectiveness of the promotions (Park and Farr, 2007; Kshetri, 2018; Meani and Paglierani, 2018). 5G and AIoT penetration can engage customers effectively, hence increasing revenue. For example, facial recognition technology utilizing AI augmented machine vision for payment of goods is slowly gaining traction. This would help decrease checkout times, decrease lost sales in physical stores and improve customer experiences.

8.4.6 Smart Cities

All of the use cases discussed above and many more pave the foundation toward forming a digital society, thereby establishing smart cities. Current infrastructure in our established cities are prone to several problems such as traffic congestion, pollution, limited public resources, etc., to mention a few. These are challenges that low- latency 5G networks with distributed intelligence can address reliably. To address this challenge, AIoT-enabled devices should be able to analyze data at a much faster rate than what is possible with current network and hardware technologies. Advancement in sensor technology, AI inference at the network edge, and machine learning are adding value toward the dream of establishing intelligent societies. Realtime data processing at the network edge using AI would allow AIoT edge devices to make decisions automatically without human intervention. For example, self-driving vehicles would need to make instant decisions in milliseconds when dealing with potential road hazards, which our current network infrastructure is unable to sustain if the data is to be transmitted to the cloud for processing and returned back to the vehicle. Such mission-critical applications are only possible by bringing the power of cloud computing near to the network’s edge. 5G distributed intelligent networks’ gigabit speed, ultra-low latency, and ultra-high reliability are essential for this effort. They would enable seamless machine-to-machine communication between AIoT sensors and devices. The potent amalgamation of edge computing and 5G would enable cities to optimize every aspect of its operations, from waste disposal management to traffic management, environmental monitoring, etc., thus enabling innovative services.

However, future systems would need to guarantee data privacy, security, and integrity and not be prone to hacking before they can be deployed in critical city infrastructures such as power plants. The end-to-end security integrated into 5G networks would allow diverse AIoT devices with customized security parameters and tolerances to be safely plugged into the network. 5G intelligent networks should allow for mMTC for transmission and reception of small data blocks over low-bandwidth pipelines for general applications such as environmental sensing, logistics, etc. At the same time, these intelligence networks should also cater to critical machine type communications for delay-intolerant, secure, and reliable transmission applications such as autonomous vehicles, healthcare, traffic control, power plants, etc. Automated intelligent management of connected resources and operations citywide can provide an efficient and cost-effective solution.

< Prev   CONTENTS   Source   Next >