Elsevier

Computer Communications

Volume 131, October 2018, Pages 77-82
Computer Communications

Reprint of: From cloud-based communications to cognition-based communications: A computing perspective

https://doi.org/10.1016/j.comcom.2018.09.006Get rights and content

Abstract

Traditional cloud-based communications provide powerful cloud computing services. However, simply supporting intensive data processing is not sufficient, especially when capacity is limited and ultra-low latency is required. Thus, it is critical to propose a new Artificial Intelligence (AI)-enabled heterogeneous networks, including various terminal networks, fogs and clouds. Derived from cognitive science and data analytics, cognitive computing can mimic or augment human intelligence. When such cognitive intelligence is integrated with communications, traditional services will be renovated with higher accuracy and lower latency. In this paper, we propose cognition-based communications, which originates from both AI-based intelligent computing and advances in communications. Then, we introduce two applications of cognition-based communications, including user-centric cognitive communications, and cognitive internet of vehicles. Through cognition-based communications, we can better meet users’ needs, provide them with a better Quality of Experience (QoE), and achieve a higher energy efficiency.

Introduction

Novel information services and applications are expanding globally with the rapid development of wireless communication and networking technologies. Advanced networks and communications can greatly enhance users’ experience and have made a huge impact in all aspects of people’s lifestyles at home, at work, in social exchanges, and economically. Although these advanced techniques have extensively improved users’ Quality of Experience (QoE) [1], they are not adequate to meet various requirements such as seamless wide-area coverage, high-capacity hot-spot, low-power massive-connections, low latency high-reliability, and other challenging scenarios. Therefore, it is critical to develop smart wireless communication and networking technologies to support optimized management, dynamic configuration, and fast service composition. Recent year have witnessed that the fusion of computing and communications exhibits a trend to reach such a goal. Cognitive computing, which is derived from cognitive science and data analytics, can mimic or augment human intelligence [2]. In addition, cognitive computing exhibits great potentials to power smart wireless communications, e.g., in self-driving. An intelligent network can be viewed as an existing network integrated with cognitive and cooperative mechanisms to promote performance and achieve intelligence. Under the new service paradigm, there are various technical challenges and problems that need to be addressed to extensively improve the user’s QoE, such as complicated decision making for routing, dynamic and context-aware network management, resource optimization, and in-depth knowledge discovery in complex environments. Artificial Intelligence (AI) plays an important role in the ability of cognitive wireless communications to meet many of these technical challenges. Furthermore, wireless communication and network ecosystems must be upgraded with new capabilities, such as the provisioning of personalized and smart Fifth Generation (5G) network services that are assisted by data cognitive intelligence, advanced wireless signal processing based on deep learning, optimized wireless communication physical layer design based on reinforcement learning, adaptive wireless resource management based on cognitive intelligence, etc.

Though various previous works used the similar terminology of “cognitive communications”, they focus on the research related with cognitive radio. For example, Green Cognitive Communications in [3], and Cognitive Device-to-Device (D2D) Communications in [4]. In order to avoid the ambiguity, our proposed architecture is named as “cognition-based communications”. In this paper, we first introduce cloud-based communications [6], including Cloud Radio Access Network (C-RAN) [7] and Name Data Network (NDN). Then, we present the architecture and applications of cognition-based communications. In summary, the main contributions of this paper include:

  • 1)

    We introduce two representative paradigms of cloud-based communications, i.e., computing-centric C-RAN, and cache-enabled NDN.

  • 2)

    We propose cognition-based communications with a new architecture that includes two layers, i.e., the communication layer and cognition layer. Specifically, the cognition layer consists of two core cognitive engines, i.e., the resource cognitive engine and data cognitive engine.

  • 3)

    The benefits of applying the proposed cognition-based communications are illustrated by two archetypal wireless networking applications, i.e., user-centric cognitive communications and cognitive internet of vehicles [5].

The remainder of this paper is organized as follows: Section 2 presents the architecture of conventional cloud-based communications, and give its pros and cons. In Section 3, the evolution from cloud-based communications to cognition-based communications is identified from a computer communications' point of view. Section 4 concludes this paper.

Section snippets

Cloud-based communications

Cloud-based communications include computation-oriented C-RAN and cache-centric NDN. As shown in Fig. 1, C-RANs offload communication-related computing onto the cloud for centralized processing and management. In NDNs, to decrease the delay of content retrieval, contents are cached in various network nodes in a distributed fashion [2].

Cognition-based communications

In this section, we will first introduce how to leverage cognition for the optimization of communications in the wireless network. Then, we give the architecture of cognitive-based communications. Finally, we provide two applications of cognition-based communications.

Conclusion

In this paper, we have summarized the status and characteristics of the traditional cloud-based communications. Then, by combining traditional cloud-based communications with current popular cognitive computing technology, we have analyzed the feasibility of evolution to cognition-based communications in wireless communication networks. We have proposed the architecture of cognition-based communications, and described the two important modules, i.e., resource cognitive engine and data cognitive

Acknowledgement

This work was supported by the National Key R&D Program of China (2018YFC1314600), the National Natural Science Foundation of China (under Grant No. U1705261, 61671088, 61572220), Director Fund of WNLO, the Fundamental Research Funds for the Central Universities (HUST: 2018KFYXKJC045), and the Canadian Natural Sciences and Engineering Research Council.

References (9)

There are more references available in the full text version of this article.

Cited by (2)

  • Resource provisioning towards OPEX optimization in horizontal edge federation

    2020, Computer Communications
    Citation Excerpt :

    Although cloud computing resources have been widely used on-demand by distributed users, cloud services are situated far away from UEs and cannot satisfy the offloaded request requirement [1,2]. Therefore, it will be an inconsistency with serving ultra-low latency request in low expenditure through tradition cloud-based services [3]. Edge computing can be addressed to serve such kind of requests [4,5].

  • Delay-Aware and Profit-Maximizing Task Migration for the Cloudlet Federation

    2022, International Journal of Advanced Computer Science and Applications
View full text