| Software Defined Radio(SDR)technology is being paid increasingly attention because of its easily deployment and low cost.By softwarization of the hardware facilities of cellular communication,the development cycle can be shortened,and an experimental platform can be provided for the verification of new technologies and network architectures.The OpenAirInterface(OAI)launched by EURECOM is a relatively perfect open-source SDR hardware platform at present.Based on the OAI platform,the concrete research on design and application of hardware demonstration platform for fog radio access network is carried out in this paper.Firstly,based on the software and hardware deployments of OAI,a 4G(4th Generation)hardware demonstration platform is constructed in this paper.Based on the OAI open-source platform,the hardware demonstration platform for fog radio access networks(F-RAN)is designed according to their architecture characteristics.The platform contains fog-computing-based access point(F-AP),fog-computing-based user equipment(F-UE),a cloud server and so on.Subsequently,the throughput of users under three different scenarios are tested based on the constructed hardware platform.The results are analyzed and compared with the theoretical peak rate.Secondly,the video traffic in cellular networks is increasing dramatically nowadays.In order to effectively meet the characteristics of high capacity and low latency of high-definition(HD)video service,the edge cache and computing capabilities of edge nodes in F-RAN should be fully utilized.The proposed architecture of HD video distribution is designed just according to the edge cache characteristic of F-RAN.The first application demonstrated on the hardware platform is HD videos distribution.On the basis of traditional cloud server-based video distribution scheme,a local video distribution strategy based on F-AP is proposed in this paper.The demonstration of two different distribution schemes proves that the F-RAN architecture can effectively reduce the transmission latency of service and alleviate the traffic burden on the capacity-constrained fronthaul.The second demonstration is image recognition and target classification based on HD video stream.In the traditional cloud-based scheme,the target classification is conducted after the video images are transmitted to the cloud server by RTSP(Real Time Streaming Protocol).In order to solve the problems of long latency and limited recognition ability of this classification scheme,an edge-based target classification scheme is proposed in this paper,whose recognition and classification process is carried out directly on local F-AP.The latency performance and recognition accuracy of the latter scheme are better than those of the cloud-based scheme,and more types of objects can be identified.At last,the demonstrations of Web AR(Web Augmented Reality)applications are conducted on the F-RAN hardware platform in this paper.Compared with the traditional in-app AR applications,Web AR has many advantages,such as good cross-platform performance,fast deployment and easy dissemination.The first Web AR application designed is Web Face AR based on Multi-task Cascaded Convolutional Neural Network(MTCNN)in this paper.It realizes the function of face detection and recognition.After successfully recognizing the face,the person's personal information will be displayed below the bounding box.The second Web Space AR application is developed based on AR.js repository.A three-dimensional rotat:ing Earth will be displayed on the screen of the mobile phone when the trained Quick Response(QR)code pattern is scanned by a webcam.Based on the open-source OAI project,the hardware demonstration platform for fog radio access network is designed and built in this article,and the advantages of F-RAN architecture have been proved through the demonstrations of typical traffic which are HD video distribution and Web AR application.This demonstration platform will also provide a strong guarantee for the later related research. |
PDF Full Text Request