Real Time Convex Optimisation for 5G Networks and Beyond
2: Queen's University Belfast, Belfast, UK
3: School of Electrical and Data Engineering, University of Technology Sydney, Sydney, Australia
There is no doubt that we are facing a wireless data explosion. Modern wireless networks need to satisfy increasing demand, but are faced with challenges such as limited spectrum, expensive resources, green communication requirements and security issues. In the age of internet of things (IoT) with massive data transfers and huge numbers of connected devices, including high-demand QoS (4G, 5G networks and beyond), signal processing is producing data sets at the gigabyte and terabyte scales. Modest-sized optimisation problems can be handled by online algorithms with fast speed processing and a huge amount of computer memory. With the rapid increase in powerful computers, more efficient algorithms and advanced parallel computing promise an enormous reduction in calculation time, solving modern optimisation problems on strict deadlines at microsecond or millisecond time scales. Finally, the interplay between machine learning and optimisation is an efficient and practical approach to optimisation in real-time applications. Real-time optimisation is becoming a reality in signal processing and wireless networks. This book considers advanced real-time optimisation methods for 5G and beyond networks. The authors discuss the fundamentals, technologies, practical questions and challenges around real-time optimisation of 5G and beyond communications, providing insights into relevant theories, models and techniques. The book should benefit a wide audience of researchers, practitioners, scientists, professors and advanced students in engineering, computer science, ubiquitous computing, information technology, and networking and communications engineering, as well as professionals in government agencies.
Inspec keywords: disasters; convex programming; embedded systems; 5G mobile communication; telecommunication computing; optimisation; autonomous aerial vehicles
Other keywords: computational complexity; embedded systems; optimisation; 5G mobile communication; convex programming; relay networks (telecommunication); emergency management; resource allocation; disasters; autonomous aerial vehicles
Subjects: Mobile radio systems; Optimisation techniques; Mobile robots; Aerospace control; General and management topics; General electrical engineering topics; Communications computing; Education and training; Optimisation techniques
- Book DOI: 10.1049/PBTE087E
- Chapter DOI: 10.1049/PBTE087E
- ISBN: 9781785619595
- e-ISBN: 9781785619601
- Page count: 223
- Format: PDF
-
Front Matter
- + Show details - Hide details
-
p.
(1)
-
1 Convexity and convex optimisation problems
- + Show details - Hide details
-
p.
1
–7
(7)
Over the past years, convex optimisation theory has proven its important role in engineering. In particularly, convex optimisation-based models are applied in numerous signal-processing and wireless communication scenarios. Together with the explosion of wireless communication, convex optimisation has become the most poten-tial approach for the design, analysis and deployment of wireless communication systems. In fact, many aspects of wireless networks such as beamforming design, resource allocation, spectral and energy efficiency maximisation are exploited and addressed by convex optimisation. In addition, many non-convex optimisation problems in wireless networks can be solved by non-convex approaches or can be converted into convex ones which are handled using various convex optimisation algorithms.
-
2 Recognition and classification of convex programming
- + Show details - Hide details
-
p.
9
–17
(9)
There are many benefits of unconstrained and constrained convex optimisation for signal processing and wireless communication. The recognition of a convex optimisation problem should be done before the method for solving the problem by optimisers can be applied. In this section, we will provide some basic approaches to recognising a convex function. A convex optimisation problem is then recognised based on the convexity of its functions.
-
3 Convex optimisation for signal processing and wireless communication
- + Show details - Hide details
-
p.
19
–24
(6)
Optimisation is often applied to solve a wide range of problems in communica-tions and signal processing such as system design, filter design, resource allocation, antenna design or, in general, any task that involves convex optimisation concepts. In each design problem, optimisation methods should be recognised and appropri-ate techniques should be selected to handle the problem. Convex optimisation has played an important role in model analysis, algorithm design and network perfor-mance optimisation. Since there are numerous scenarios of convex optimisa-tion in wireless communication, let us briefly introduce some specific aspects of wireless communication applications that use optimisation in recent years.
-
4 Introduction to real-time embedded optimisation programming
- + Show details - Hide details
-
p.
25
–35
(11)
In this chapter, we review major aspects of real-time embedded optimisation programming, such as the concept and program structure of real-time operation, timing complexity analysis and specification frameworks for real-time embedded systems. Our analysis gives a comprehensive overview of real-time optimisation programming and demonstrates the effectiveness and applicability of the real-time optimisation approach for the design of engineering systems.
-
5 Introduction to practical optimisation problems
- + Show details - Hide details
-
p.
37
–50
(14)
Optimisation problems are essential to engineering. Each optimisation application is the balanced trade-off of logistical and design strategies. Currently, there is a huge gap between optimisation theory and optimisation implementation in practice, resulting from the lack of research into this area.
-
6 First-order methods for real-time optimisation
- + Show details - Hide details
-
p.
51
–63
(13)
Although gradient-based methods are used widely in real-time optimisation, they may not be efficient when the gradient and Hessian information is difficult to evaluate, e.g., there are no explicit function forms or non-differentiable functions. To overcome this issue, derivative-free methods such as evolutionary algorithms or swarm intelligence-based algorithms can be used for finding the optimum from potential solution candidates. Otherwise, higher-order approaches based on higher-derivative information such as second-order algorithms are appropriate when high-accuracy performance is required, i.e., the exact optimum of the problem is necessary. As for higher-order algorithms, the number of derivative operations or the evaluation of costs is also largely depending on the problem size in large-scale, big-data analysis or multi-objective optimisation. Higher derivatives are often difficult to evaluate, even in second-order operation for complex objectives. On the other hand, while zero-order algorithms can achieve the exact optimal solution, they often converge very slowly without guaranteeing the exact optimal solution and convergence rate. In summary, with roots in derivative context and convexity optimisation programming, first-order methods are the most popular in the context of optimisation.
-
7 Distributed and parallel computing for real-time optimisation
- + Show details - Hide details
-
p.
65
–75
(11)
In future wireless communication, various complex and interrelated problems may concur, albeit within a temporal sequence. Large-scale problems, big-data analysis, real-time predictive control, multi-objective problems and hybrid resource allocation of communication systems are some good examples. In this chapter, we have several reasons to discuss distributed approaches and parallel computing for optimisation of wireless networks, especially for real-time applications. Distributed and parallel computing is better suited for the modelling, simulation and understanding of complex, real-world problems. From the functional point of view, it is only a little difference between the implementation with a centralised architecture and that with a distributed architecture. However, a number of arguments are in favour of the distributed approach for the implementation of hard real-time systems. Two important aspects of distributed or parallel computing architecture that make it suitable for the real-time implementation of a large-scale system are composability and scalability. In a composable architecture, the properties of the main system follow its subsystems' properties. In real-time systems, the communication interface between the host computer in a node and the communication network is fully required in both the value domain and time domain. Scalable architecture requires the unlimited extensibility of the distributed system. In other words, the complexity of system operation should be independent of the system size. In this chapter, we start with an overview of distributed computing platforms for real-time system architecture. The state messages facilitate the exchange of state information among the interconnected nodes within a communication network and enforce the autonomy of the nodes.
-
8 Machine learning for real-time optimisation
- + Show details - Hide details
-
p.
77
–89
(13)
With the boundless quality-of-service (QoS) requirements of 5G networks and beyond, communication systems must be more dynamic and intelligent and satisfy many network demands simultaneously. The concepts of artificial intelligence (AI), machine learning (ML), deep learning (DL), have found their applications in wireless communications. As wireless communication systems in dynamic environments rapidly change over time, more unexpected behaviour patterns and complicated scenarios will develop. Fortunately, ML can use robust algorithms to calibrate itself to newly acquired knowledge, provide low-complexity estimates for system model, support self-organising systems with limited human intervention. For instance, in resource allocation optimisation problems, ML attempts to reduce the complexity of optimisation problems by shrinking the solution space using feature selection technique and employing meta-heuristic solution methods for multi-objective optimisation such as finding initial solutions or choosing appropriate heuristic methods.
-
9 Real-time embedded convex programming
- + Show details - Hide details
-
p.
91
–94
(4)
A diagram of an embedded convex optimisation system is presented. As illustrated, a practical optimisation problem is the input of the system. Efficient methods are used for solving the problem on a computer (central processing unit (CPU), graphics processing unit (GPU), Field Programmable Gate Array (FPGA)), applying novel approaches for real-time optimisation, e.g., first-order methods, parallel approaches and learning-based optimisation algorithm. A custom code is produced via manual or automatic code generation by an embedded software before being integrated into the real system. There are many optimisation software packages available for both research and practical use. In this chapter, we will introduce some free optimisation tools that can be exploited for real-time optimisation applications.
-
10 Real-time embedded optimisation in UAV communications
- + Show details - Hide details
-
p.
95
–101
(7)
Unmanned aerial vehicles (UAVs) have been emerging to become a major trend in the next generation of wireless networks. With flexible configuration and mobility, UAVs can be more efficient and inexpensive for deployment, which is critical in Internet-of-Thing (IoT) applications. UAVs can easily gather information, manipulate physical objects or engage some equipment in remote or dangerous places. With a wide variety of vehicle types, UAVs can operate not only on the ground but also in a variety of environments such as space, air, water or underground. As a result, there are many types of applications where UAVs can be exploited such as environmental remediation, navigation in order to gather data, military applications, transportation of goods and performing dangerous tasks. For instance, sensor nodes can be deployed with UAVs to estimate the path and velocity of tracked vehicles. For real-time applications, the tracked results will be collected by the UAVs and reported to the central system within strict time deadlines
-
11 An introduction of real-time embedded optimisation programming for UAV systems
- + Show details - Hide details
-
p.
103
–119
(17)
For disaster communications, it is very challenging for the contemporary wireless technology and infrastructure to meet the demands for connectivity. Modern wireless networks should be developed to satisfy the increasing demand for quality-of service (QoS) in mission-critical communications for disaster management, which are currently faced with the challenges of limited spectrum, expensive resources, reliable and green communication. There is a tremendous need for optimisation techniques in the study and design of the key functionalities of wireless systems. Until now, almost all current optimisations are often carried out on large timescales (e.g., minutes or hours) without strict time constraints for solving the problems. With the improvement of computational speed, efficient algorithms and advanced coding approaches, a framework of real-time optimisation programming, which plays a major role in the trend of modern engineering such as mission-critical communications, is introduced in the context of natural disaster. In particular, this chapter gives an introduction of embedded convex optimisation programming for unmanned aerial vehicle (UAV) communications in disaster networks with strict supervision on execution time in real-time scenarios.
-
12 Real-time optimal resource allocation for embedded UAV communication systems
- + Show details - Hide details
-
p.
121
–129
(9)
This chapter considers device-to-device (D2D) wireless information and power transfer systems using an unmanned aerial vehicle (UAV) as a relay node. As the energy capacity and flight time of UAVs are limited, a significant issue in deploying the UAV is to manage energy consumption in real-time application, which is pro-portional to the UAV's transmit power. To tackle this important issue, this chapter develops a real-time resource allocation algorithm for maximising the energy efficiency (EE) by jointly optimising the energy-harvesting time and power control for the considered D2D communication embedded with the UAV. This chapter demonstrates the effectiveness of the proposed algorithms as running time for solving them can be conducted in milliseconds.
-
13 Real-time deployment and resource allocation for distributed UAV systems in disaster relief
- + Show details - Hide details
-
p.
131
–141
(11)
This chapter provides a robust and efficient resource allocation for embedded UAV-enabled cellular networks in disaster communications. To recover network in disaster area, a fast user (UE) clustering based on K-means procedure and distributed control power coefficient will be proposed and can be embedded programming in the real system by using UAV-assisted relaying for real-time recovering and rescuing working network during and after disasters. The algorithms of low computational complexity with fast convergence are proposed for our expected solution. Numerical examples are provided to demonstrate the benefit of the proposed computational approaches.
-
14 Practical optimisation of path planning and completion time of data collection for UAV-enabled disaster communications
- + Show details - Hide details
-
p.
143
–154
(12)
This chapter proposes efficient optimisation methods for embedded relay-assisted unmanned ariel vehicles (UAVs) in wireless sensor networks (WSNs) to cope with the hazardous effect of natural disaster. Particularly, by using advanced optimisation techniques, proposed low-complexity procedures are suitably applied to internet-of-things (IoT) applications when the execution time is strictly governed in disaster scenarios. Our model considers real-time optimisation in embedded UAV-WSN communication for tracking and gathering sensor data. The algorithms have low computational complexity with fast deployment and low execution time for solving our problem in milliseconds. Numerical results are shown to demonstrate the benefit of our proposed approaches for UAV-WSN.
-
15 Learning-aided real-time performance optimisation of cognitive UAV-assisted disaster communication
- + Show details - Hide details
-
p.
155
–167
(13)
This chapter proposes efficient optimisation methods for relay-assisted unmanned aerial vehicles (UAVs) in cognitive radio networks (CRNs) to cope with the network destruction in the event of a natural disaster. The model considers real-time optimisation in embedded UAV-CRN communication invoked for recovering wireless communication services. Particularly, by conceiving advanced optimisation techniques and training deep neural networks (DNNs), proposed solutions become capable of supporting real-time applications in disaster recovery scenarios. The algorithms impose low computational complexity, hence, have a low execution time in solving real-time optimisation problems. Numerical results demonstrate the benefits of our approaches proposed for UAV-CRN.
-
References
- + Show details - Hide details
-
p.
169
–188
(20)
-
Appendices
- + Show details - Hide details
-
p.
189
–199
(11)
-
Back Matter
- + Show details - Hide details
-
p.
(1)