Abstract

   As the coming maturity of Internet of Things, many home-networked devices with various sensors are deployed in residential environments. It is important to accommodate many devices that compete for bandwidth allocation to transmit data inwards and outwards through a certain home network domain. This paper proposes a novel mechanism which can adjust media stream qualities and transmission bandwidth allocations to different devices. This mechanism is agile against dynamic changes of data workload and finite network resources between inside and outside of a home network. Practical demonstration exhibits that the effects are able to maintain differentiated media transfer services in IoT-based home networks.

Chao-Yu Hsu*, Chih-Lin Hu*, Kun-Sheng Huang*, Yung-Hui Chen**, Jiun-Long Huang***

(*National Central Univ., Taiwan; **LungHwa Univ. of Science and Technology, Taiwan; ***National Chiao Tung Univ., Taiwan )

Abstract

   Distribution of medicine in the public health community centers has been regulated and monitored in an official distribution channel. However, counterfeit medications still circulate in the community. Circulation of counterfeit medicines is caused by factors from businesses, consumers, and medicine distribution supervision. The use of blockchain technology in supply-chain has shown new breakthroughs to overcome these problems since it ensures immutability, transparency, and decentralization of data. In this paper, we proposed a new blockchain-based Medicine Supply Chain Management (MSCM) system. The proposed system has been applied to medicine distribution channels between public community health centers in Indonesia. To evaluate the proposed system, we used JMeter as a benchmarking tool to measure the system performance in terms of transactions per second, latency, and the utilization of resources.

Hartanto Kumiawan, Jung Yeon Kim, Hong Taek Ju (Keimyung Univ., Korea (South) )

Abstract

   Effective network congestion control strategies are the key to secure the normal operation of complex and changeable networks. The fundamental assumptions of many existing TCP congestion control variants dominated by hand-crafted heuristic algorithms are no longer valid. We propose an algorithm called TCP-Proximal Policy Congestion Control (TCP-PPCC), which is based on deep reinforcement learning algorithm Proximal Policy Optimization (PPO). TCP-PPCC updates the policy offline from the features of the preceding network state and feedback from the current network environment and adjusts the congestion window online with the updated policy. The senders with TCP-PPCC can learn about the changes in network bandwidth more accurately and adjust the congestion window in time. We demonstrate the performance of TCP-PPCC by comparing it with the traditional congestion control algorithm NewReno in four network scenarios with the ns-3 simulator. The results show that in scenario 2, TCPPPCC takes 58.75% improvement in average delay and 27.80% improvement in throughput compared with NewReno.

Jing Li, Yuyao Guan, Pengpeng Ding, Shiwei Wang (Shanghai Univ. of Electric Power, China )

Abstract

   Internet of Things (IoT) integrates with finger tracking and 3D signature verification can provide the user authentication service for these IoT applications which require user identification and authorization, such as the home automation and security system. This work develops two modules. The FingerTalk module provides the fingertip trajectories tracking service based on a Leap Motion camera and IoTtalk. The AirSig module provides 3D signature verification service to recognize whether a signature is signed form the authenticated user. The AirSig module requires only two user signature samples to train the neural network model which significantly reduces the process time for signature collection. The validation results from 10 people show the average accuracy is up to 93.6% on 3D signature verification.

Yun-Wei Lin, Lin-Hang Kao, Chun-Ting Yeh (National Chiao Tung Univ., Taiwan )

Abstract

   This paper is to improve the performance of Hyperledger Fabric that is one of the private blockchain opensource projects. The decentralization of data management, which has recently become important, is realized as a blockchain. Blockchain is a technology used in various fields such as data verification, storage, and banking, but performance is one of the important problems and is always mentioned. This is because a transaction must be agreed quickly and recorded in the ledger so that it can be used smoothly in systems that generate tens of thousands of transactions per second, such as banking and card use. Hyperledger fabric sends transactions generated by clients to peer nodes on the network. The transactions are hashed with the previous block in the distributed ledger. Peer nodes collect transactions received from multiple clients and execute chain code for verification, which is then sent to the orderer node and made into blocks. The block created at this time compares hash values to several peer nodes to verify the integrity of the data after verifying the block. Orderer node collects the transactions and is ordering them into the block. In this paper, we proposed to improve the performance of the hyperledger fabric by reducing network latency and the number of transfers by batch processing the process of sending and receiving chain code messages in the process of delivering transactions generated by clients for verification. The results of the performance are expressed in transactions per second(TPS), with the experiment showing that send rate(TPS) nearly doubling to 2302→4244, and throughput(TPS) increasing by about 1.5 times to 1677→2244. As a result of the experiment, send rate and throughput increased by a large margin, the batch processing of chain code messages can perform much higher performance.

JungWon Lee, Park Sejin (Keimyung Univ., South Korea )

Abstract

   Since the development of Bitcoin, the first blockchain-based cryptocurrency, many cryptocurrencies have formed and have traded in markets. The integrity and anonymity of cryptocurrency was enough to raise its value and its price gained worldwide attention. Therefore, many studies are being carried out to predict the price of cryptocurrency for make a profit. We cluster time series through K-Medoids algorithm and train and evaluate each cluster with predictive models. We also examine the predictive performance in Bitcoin price according to the various distance measurement of clustering.

Ui-Jun Baek, Mugon Shin, Min-Seong Lee, Boseon Kim, Jee Tae Park, Myung-Sup Kim (Korea Univ., South Korea )

Abstract

   The Business-to-Business-to-X (B2B2X) model has increased the importance of orchestrators that build and operate services that consists of various wholesale services. To develop RESTful API-based services, service providers need to catch up on the specifications of new wholesale services and specification changes of existing services quickly and inexpensively. Therefore, it is important to develop software called an API adapter that absorbs API differences from various services quickly and inexpensively. In this paper, we propose a method for comprehensively automating testing of not only C-Plane signals such as service provisioning/change/abolishment but also U-Plane signals such as communication between user devices and servers. We implemented our proposal using open source software and found that it improved test efficiency in an actual software development project.

Sho Kanemaru, Tomoki Ikegaya, Kensuke Takahashi, Tsuyoshi Toyoshima (NTT Corporation, Japan )

Abstract

   As for the service assurance operation of telecommunications carriers, accurate information about a network topology which indicates the connection relationships between pieces of network equipment is necessary. However, the network of a telecommunications carrier has several hundreds of thousands of equipment, and its topology is frequently supplemented and modified due to daily construction work and troubleshooting. It is a therefore a problem when incorrect topology information is mixed into the overall topology information. In this paper, we propose a method that can discover the topology between equipment by using alarm information issued by those equipment during construction work or when a failure occurs. The proposed method was evaluated using alarm information generated under in certain commercial configurations (sections containing specific routers), and it was confirmed that the current topology could be discovered with 100% accuracy, even though only 1.7% of the total topology was evaluated in one day.

Atsushi Takada, Naoki Hayashi, Mizuto Nakamura, Naoyuki Tanji, Toshihiko Seki, Kyoko Yamagoe (NTT Corporation, Japan )

Abstract

   To support network slicing and service function chaining (SFC) at a time, we propose a novel hierarchical identifier (HID)-based 5G architecture. For this, we first introduce HID which consists of network slice selection assistance information (NSSAI) and service path ID (SPI). Based on HID, a user can attach a specific network slice and flows generated by the user can be processed by a set of service functions (SFs) in a sequence. Meanwhile, when the incoming flow to a specific slice unexpectedly increases, the slice cannot handle incoming flow due to its limited capacity, which degrades users’ quality of service (QoS). To alleviate this issue and efficiently utilize network resources, we introduce a concept of the backup slice shared by different services. Evaluation results demonstrate that the proposed architecture can achieve better performance in terms of the average system blocking probability and utilization in dynamic environments.

Haneul Ko, Jaewook Lee, Hongrok Choi, Sangheon Pack (Korea Univ., South Korea )

Abstract

   At existing cloud computing environment, virtual machine allocation mechanisms have been actively applied to resources management for processing dynamic workload in largescale data center. In this paper, we address the problem of how many virtual machines should be allocated to a server, so that a given percentile of the execution time of a job is bounded by a predefined value. We first consider the case of “dedicated CPU”, whereby we calculate the CPU allocated to a single VM. Subsequently, we extend the analysis to the case where the CPU is shared equally by different groups of VMs. In this case, we calculate how many VMs need to be allocated in each group.

Yue Han*, Harry Perros**, Mingwu Yao***

(*National Univ. of Defense Technology, China; **North Carolina State Univ., USA; ***Xidian Univ., China )

Abstract

   More and more telecom operators around the world began to build and to deploy fifth-generation (5G) mobile networks. With the advent of 5G service scenarios, it brings new challenges and increases the complexity of management. To deliver good quality of experience of these 5G applications for customers, service assurance is important. In this paper, we propose an end-to-end service assurance solution for 5G crosshaul networks. The 5G crosshaul network architecture consists of fronthaul, midhaul, and backhaul networks. Our solution deployed in 5G trial network in Chunghwa Telecom in Taiwan. The solution focus on slicing service assurance and synchronization monitoring. Besides, our solution adjusts most of network management in the operation support system (OSS) to apply to 5G crosshaul network. Our goal is to provide stable 5G transport network and to assure quality of experience for customers.

Che-Chun Teng, Chen Mei-Chun, Min-Han Hung, Hong-Jie Chen

(Chunghwa Telecom Co., LTD., Taiwan )

Abstract

   The cable TV network is considered as an optimal network for convergence of broadcasting and communication because it is possible to provide two-way services. Recently, the cable broadcasting network is in shortage of available frequency resources due to high quality broadcasting services, increased demand for VoD, and increased bandwidth of high-speed Internet services. This paper presents a new method for switched digital video (SDV) technology to obtain additional transmission resources in cable broadcasting networks. In the proposed SDV method, communication and broadcasting service share the same frequency band, it is possible to improve the frequency usage efficiency by providing switching of broadcasting channels without complex system construction.

Joon-Young Jung (ETRI, South Korea )

Abstract

   With the development of various technologies such as high-speed Internet and SNS dissemination, there have been many fields that require processing of big data generated in real time. Accordingly, real-time streaming data processing technology has been developed, and representative platforms include Apache Storm, Apache Spark, and Hadoop. These processing technologies provide scalability to configure distributed systems using multiple servers because they vary in performance, such as throughput and processing speed, depending on the server environment, but the more the number of servers, the more difficult it is to manage. To solve this problem, a problem can be solved by using a docker, a kind of virtualization system that provides ease of expansion. However, there is a place to maintain a native environment without using Docker due to the problem that performance may be reduced, which is a disadvantage of all virtualization systems. In this paper, we build Apache Storm and Apache Spark, which are real-time data processing systems in Docker and Native environments and conduct performance measurements through experiments processing JSON-format data to verify how much performance decreases in Docker environments.

Jiwon Bang, Mi-Jung Choi (Kangwon National Univ., South Korea )

Abstract

   Bitcoin users are guaranteed to be anonymous, increasing the number of cryptocurrency trading related to crimes and fraudulent activities. While most studies about detecting illegal transactions try to distinguish trading patterns and classify them from legitimate ones, classification performance is poor since the class distributions of transaction data are highly imbalanced. In general, the Synthetic Minority Over-sampling TEchnique (SMOTE) is used to deal with classimbalanced data, but SMOTE has a problem that it does not fully represent the diversity of the data. In this paper, we introduce another oversampling technique using Generative Adversarial Networks (GAN) to generate artificial training data for classification model. In order to verify similarity between artificial data and the actual one, oversampled dataset is evaluated with a classification model using XGBoost algorithm. We show classification performance is improved on average with synthetic data generated by both SMOTE and well-designed GAN model.

Jungsu Han, Jongsoo Woo, James W. Hong (POSTECH, Korea (South) )

Abstract

   This research provides a method for integrated centralized monitoring under the 5G Crosshaul architecture. Our method establishes a process to provide integrated monitoring technology and intelligent obstacle location for endto-end circuits across multiple domains. The monitor page shows the operating status of the 5G Crosshaul service composed of Multi-Technology, which are a base station, Mobile Fronthaul (MFH), Mobile Backhaul (MBH) and core network. The process proposed by this method includes the following technologies, encapsulating and formalizing cross-domain services and service analysis mechanisms. For the service, it integrates the service status and abnormal events of network equipment and circuits through regularized data to integrate data from different domains. The service-impacted level based on the importance of resources and provide a reference to the basis of the handling of obstacles. In addition, provide a notification mechanism to proactively notify relevant personnel, and show root cause analysis result. Through the above methods, providing Root cause of obstacles and abnormal impact range of services, reduce the time for removing obstacles and improve service quality.

Ting-Che Chuang, Yueh-Hsien Lin, Chien-Wen Yang, Min Liu, Yong-Zen Haung (Chunghwa Telecom Co., LTD., Taiwan )

Abstract

   With the development of cloud technology, people are focusing on lightweight container technology. Container networking is a core technology in providing high-level cloud services, and there are various implementation models such as Container Network Model (CNM) and Container Network Interface (CNI). Among them, CNI is a de-facto standard adopted by various container platforms. There are many plugins that implement the CNI networking model, and the implementation method differs, showing a great difference in a view of performance. MEC (Multi-Access Edge Computing) is a technology that provides services by locating the server closest to the user who wants to use mobile communication services. In this paper, the performance was measured by applying various CNI network plugins to the CoV architecture. Through this measurement, we analyzed the factors of network performance degradation. The result of the analysis is expected to be used as a good reference in constructing the CoV architecture for low latency in the future.

Junghan Yoon*, Jian Li**, Sangho Shin** (*KAIST, South Korea; **SK Telecom, South Korea )

Abstract

   To solve the issues of high communication cost and slow information transmission in the applications of agents’ collaboration, in the paper, we propose an approach to the construction of dynamic small-world network based on attention mechanism. Small-world model is applied to handle information exchange among agents. During the movement to destination, agent obtains individual historical optimal position through self-attention mechanism, and obtains local optimal position of k-neighbor through local attention mechanism, enabling search for the next optimal position. The experimental results verify effectiveness of our method, as well as stability and invulnerability which can adapt to dynamic changes.

Rong Xie (Wuhan Univ., China )

Abstract

   In this paper, we present an IEEE 802.11ah (HaLow) dongle that provides sub-1GHz wireless channel and USB (universal serial bus) interface for power supply and wired data link for easy Internet of Things (IoT) wireless networking. The HaLow dongle is implemented using CC1352P1 sub-1GHz smart RF chipset that provides 50Kbps ~ 4Mbps in 10m ~ 1Km range. The major advantages of the HaLow dongle include i) easy installation using USB interface, ii) energy efficient wireless networking with flexible channel selection and Tx rate/power adjustments, and iii) user-friendly API (application programming interface) for raw-socket-based application developments. The functional architecture of IEEE 802.11ah (HaLow) dongle is explained in detail and the performance is analyzed in practical environments.

Min-Cheol Kim, Young-Tak Kim (Yeungnam Univ., South Korea )

Abstract

   Multisensory IoT system contains multiple types of sensors, actuators, and computation devices. In this kind of system, providing different viewpoints of observations by heterogenous sensors such as RGB camera and wearable sensors can improve the performance of intelligent systems. The issues of building this system are that we have many components running on different platforms and implementing with different programming languages. It causes the complexity of architecture design, interoperability of components, and portability and also limits the system scalability. Therefore, we propose a reference design which is based on service-oriented architecture for alleviating these issues. We also present a case study of system prototyping which is constructed by adopting an RGB camera and smart insoles to create an identity recognition system.

Sevendi Eldrige Rifki Poluan, Yan-Ann Chen (Yuan Ze Univ., Taiwan )

Abstract

   To achieve high availability of traffic and maximum coverage in the 5G network, clusters of small cells approach to maximize the system capacity. In this approach, the Core Network (CN) encounters a large number of signals due to the frequent hand-off between small cells in the event of an increase in the speed of user movement and users within certain limits of small cells. The present 5G system executes all handoff signals in the traditional complex hierarchical sequence which leads the system toward the bad performance due to redundant signaling. NO Stack architecture has emerged as a promising methodology for radio access network (RAN) in the context of reducing the redundant signaling during X2-based handover by using control logic centrally and flat protocols. This paper proposes a system that is based on NO Stack architecture for reducing handoff requests in CN by assigning the dedicated mobility controller to each cluster of small cells for locally managing the mobility. The dedicated cluster controller specially designs to manage and control the cluster of small cells as well as maintain the forwarding information of user equipment locally. In this proposed system, the dedicated controller resides at the nearest edge cloud, therefore a distributed edge cloud network is developed for high computing infrastructure and low latency access for the X2-based handover request.

Adeel Rafiq, Muhammad Afaq, Wang-Cheol Song (Jeju National Univ., Korea (South) )

Abstract

   We proposes a backbone network resource optimization algorithm for video communications that use edge computing. In the current video communication architecture, the key component of video communication, called the multi-point control unit (MCU), is deployed in the central cloud server, and its bandwidth consumption in the backbone network becomes enormous as the video resolution and the frequency of use increase. By implementing edge computing, the MCU can be deployed at the entrance node of the backbone network. This allows (i) a local loopback of video traffic at an edge, and (ii) traffic compression (e.g., thumbnailing) between edge nodes. Though these characteristics can reduce the resource consumption of the backbone network, the edge deployment and routing (EDR) problem, classified as NP-hard, should be solved to sufficiently reduce the bandwidth. To solve the NP-hard EDR problem within a feasible period, we propose a divide and merge algorithm based on the linear programming approach. With our algorithm, bandwidth consumption using edge computing is reduced by approximately 30% compared with the current video communication architecture in the world-wide network.

Kouichi Genda*, Mitsuru Abe**, Shohei Kamamura***

(*Nihon Univ., Japan; **NTT Communications Corporation, Japan; ***NTT Corporation, Japan )

Abstract

   Traffic classification (TC) has a principal function in autonomous network management. Recently, deep learning and machine learning-based TC have become popular than the traditional port-based and protocol-based TC due to practices such as port disguise and payload encryption. The flow-based TC is reliable as it relies on time-related statistical features. Federated learning is a distributed machine learning technique to train improvised deep/machine learning models with less privacy distress. The organizations or enterprises having similar business models may take participation in building a federated model for their network traffic characterization. In this study, we build a cross-silo horizontal federated model for TC using flow-based time-related features. The federated model shows comparable performance to the centralized model.

Umer Majeed, Latif Ullah Khan, Choong Seon Hong (Kyung Hee Univ., Korea (South) )

Abstract

   In recent years, IP surveillance networks are expected to enable various practical applications, such as finding suspects, monitoring pedestrians, and securing societies (e.g., securing a city, a company and a data center). With these applications, IP surveillance network is regarded as one of the potential technologies toward developing smart cities. To support the concept of IP surveillance networks, automatic attribute recognition systems have emerged as a promising intelligent management system. To automatically recognize attributes of pedestrians (e.g., gender and clothing), we apply deep convolutional neural networks (CNNs), and the main contributions of this paper are threefold: (1) we proposed a practical system architecture for intelligent management of surveillance networks; (2) we implemented different deep CNNs, and an ensemble-learning method that leverages these multiple deep-learning models; (3) we evaluated the models using the real data of IP surveillance networks.

Chien-Chi Kao, Yung-Chang Lai, Jung Pei, Chih-Wei Chang, Fei-Hua Kuo, Jin-Yuan Shun (Chunghwa Telecom, Taiwan )

Abstract

   Recent years have shown a remarkable interest in federated learning from researchers to make several Internet of Things applications smart. Although, federated learning offers users’ privacy preservation, it has communication resources optimization challenge. In this paper, we consider federated learning for cellular networks. We formulate an optimization problem to jointly minimizes latency and effect of loss in federated learning model accuracy due to channel uncertainties. We decompose the main optimization problem into two sub-problems: resource allocation and device association sub-problems, due to the NPhard nature of the main optimization problem. To solve these sub-problems, we propose an iterative approach which further uses efficient heuristic algorithms for resource blocks allocation and device association. Finally, we provide numerical results for the validation of our proposed scheme.

Latif Ullah Khan, Umer Majeed, Choong Seon Hong (Kyung Hee Univ., Korea (South) )

Abstract

   Unmanned Ariel Vehicle(UAV) has been widely used in the edge computing network. Owing to its line-of-sight communication ability and mobility, it has offered task offloading service and some task computation for mobile devices(MDs). Nowadays, as high-intensity calculation application develops rapidly, maximizing the size of offloaded tasks is necessary for meeting the users’ experience in applications. In this context, our paper takes research on how to deploy UAVs at the most proper place to offer task offloading with TDMA protocol. Specifically, we will optimize the task offloading number in the MDs-UAV system by combining the location of UAVs, the task computation capacity of UAV, and the MDs-UAV associations in a certain time. We prove that the joint UAV deployment and task computation problem is NP-hard, and use a greedy algorithm to optimize the result. Compared with randomly selected method, our simulation shows that the greedy algorithm performs greatly and the deployment of UAVs are sensible. We believe that this idea will improve the system tasks processing rate and Quality of Experience(QoE) of mobile users.

Yuqing Chen, Zhaohui Zheng (Soochow Univ., China )

Abstract

   To improve throughput, 5G networks adopts beamforming technique and have high small-cell-density deployment. In Time Division Duplex (TDD) based 5G networks, Almost Blank Subframes (ABS) scheme and Cell Range Expansion (CRE) scheme are adopted to reduce inter-cell interference and to achieve inter-cell load balancing, respectively. However, existing ABS allocation and CRE are cell-based, i.e., requires that all beams inside a cell adopting the same ABS allocation. Due to the multiple overlapping of small cells, such cell-based ABS/CRE may much reduce the throughput. In this paper, we proposed a beam-based ABS allocation and beam-based CRE. To further reduce the interference, in this paper, small cell center UEs and small cell edge UEs (those UEs connects to small cell due to CRE technique) have distinct downlink schedules. According to our simulation results, our scheme have low computation cost since the computation cost of beam-based is much lower than cell-based.

Guey-Yun Chang (National Central Univ., Taiwan )