previous day
next day
all days

View: session overviewtalk overview

09:00-10:30 Session 6: Keynote Day 2: Convergence of Broadcast and Broadband

Keynote: Convergence of Broadcast and Broadband

Mark Aitken, Parag Naik, Jae-young Lee, Jerald Fritz, David Gomez Barquero

ATSC 3.0 Installation Plan and Service Scenarios in USA

ABSTRACT. As US Broadcasters start their deployments of ATSC 3.0, the conversations regarding “business models” and “use cases” becomes more focused. These future offerings begin to drive critical focus on deployment topologies and receiver requirements. An area of new focus is the densification of SFN sites, in some cases mimicking low-band LTE deployments. The impact of such deployments gives rise to frequency reuse opportunities. 

5G Offload

ABSTRACT. What is Convergence?

Saankhya continues to pioneer new “5G Convergence” migration paths, bringing innovative solutions to offload video from the congested cellular network, seamlessly and selectively using broadcast spectrum.   As video content increases dramatically, unicast mobile networks are choked resulting in a low quality of user experience (UX). Unicast - one-to-one - mobile networks were never designed to carry video broadcast traffic. Broadcasting was and is! Beyond linear video services, broadcasting’s one-to-infinite architecture to deliver simultaneous enormous amounts of data for cached and prepositioned consumption opens huge new opportunities in new technology sectors including autonomous vehicles and the Internet of Things. 

How will we implement it?

The intention is to use the UHF broadcast spectrum to deliver data - including video content - seamlessly over a platform that now uses the same Internet Protocol as the other telecom platforms.  A favorable Government policy of accepting TRAI’s recommendation will open UHF spectrum to the private sector.  The cellularized broadcast architecture designed by Saankhya will allow 3 times the spectrum reuse compared to the current 90-year-old topology of high TV tower transmission.

A one-way broadcast overlay network is implemented alongside the two-way mobile network. Network slicing – the ability to share multiple networks using the same physical infrastructure – is enhanced by the addition of the broadcast platform, allowing convergence of data, co-existence of platforms and efficient interworking all in a virtualized master network.   This integration of the broadcast network into the telecommunications Common Core, along with an analytics engine that decides when and how to offload traffic to either of the networks, yields a converged infrastructure “as a platform.”  The virtualized combined network can be optimized to offer both two-way content delivery AND linear content as well. The following picture shows how content is intelligently routed to the user by the best means – broadcast or broadband.

LDM/SHVC-base Convergence of Broadcast and Broadband

ABSTRACT. The presentation focuses on the recent research activities related to the convergence of broadcast and broadband networks. The possible service scenarios and benefits of ATSC 3.0 and 5G convergence are discussed. The recent field trials in Jeju using the LDM/SHVC broadcast cooperating with mobile network is also introduced.

ATSC 3.0 Broadcast and Broadband Convergence in USA

ABSTRACT. Business use cases for broadcasters in a 5G world – The opportunities for broadcasters to expand their services to consumers and businesses are enhanced by an order of magnitude with the ATSC 3.0 transmission standard.  It allows manipulation of the enormous data pipe to deliver not only traditional linear television entertainment, but also a host of new services that sync up naturally with telecom carriers services provided on 5G platforms.

5G-Xcast – Enabling Point-to-Multipoint Communication Capabilities in 5G New Radio and the 5G Core

ABSTRACT. The talk will provide an overview about the current status and next steps of multicast/broadcast standardization in 3GPP for 5G, covering both the LTE-based terrestrial broadcast solution enTV in Release-16 as well as stand-alone 5G in Release-16 and beyond. The talk will also present the final technical solution of the 5G-Xcast project to introduce multicast and broadcast communication capabilities in the 5G New Radio and the 5G service-enabled Core in a transparent way from end users and content providers. The 5G-Xcast solution covers air interface, Radio Access Network (RAN) architecture and core network. The proposed solution meets the requirements of public service broadcasters.

10:50-12:30 Session 7A: 5G-Xcast


5G Mixed Mode: An Innovative Point-to-Multipoint Solution for New Radio

ABSTRACT. This work presents a potential solution for enabling the use of multicast in the 3GPP Release 15 air interface, called 5G New Radio (NR). The proposed multicast mode, denoted as 5G Mixed Mode follows one of the two approaches envisaged in 3GPP, which enables a dynamic and seamless switching between unicast and multicast, both in the downlink and the uplink. This paper also provides a performance evaluation in Single Frequency Networks (SFN) and mobility scenarios, showcasing the potential advantages of this solution over unicast in relevant scenarios.

IMT-2020 Key Performance Indicators: Evaluation and Extension Towards 5G New Radio Point-to-Multipoint
PRESENTER: Jose Luis Carcel

ABSTRACT. 5G New Radio (5G NR) is the first technology standard to enable mobile communications in three different usage scenarios: enhanced Mobile Broadband (eMBB), Ultra-Reliable Low Latency Communications (URLLC) and massive Machine Type Communications (mMTC). As a result of its technical impact, ITU-R considers 5G NR as candidate technology for the International Mobile Telecommunications for 2020 (IMT-2020) evaluation process. To assess 5G NR performance, IMT-2020 defines multiple Key Performance Indicators (KPIs) related to specific requirements to enable Point-to-Point communications (PTP). IMT-2020 does not consider Point-to-Multipoint (PTM) communications due to the absence of multicast/broadcast capabilities in the current 5G NR standardized solution. Nevertheless, PTM communications are proved to be a key feature for 5G applications in a wide number of vertical sectors as Media & Entertainment, Public Warning or Internet of Things. This paper performs an analysis of different IMT-2020 KPIs for 5G NR PTP and extends the evaluation to a PTM perspective. The results contribute to enhance the role of PTM communications towards the IMT-2020 landmark.

Performance Evaluation of New Radio PDCCH for Point-to-Multipoint scenarios

ABSTRACT. This extended abstract provides a first set of the performance evaluation of New Radio (NR) Physical Downlink Control Channel (PDCCH) for Point-to-Multipoint (PTM) use cases, e.g., eMBMS (evolved Multimedia Broadcast Multicast Services). The evaluation methodology is based on analyses and link level simulations where the channel modelling is modified to model different PTM scenarios. The final version of this work aims to provide insightful guidelines on the delay/echo tolerance of the NR PDCCH in terms of mobility and coverage.

Enabling SFN Transmissions in 5G Cloud-RAN Deployments

ABSTRACT. In this paper, an innovative Radio Access Network (RAN) architecture design to enable Single Frequency Network (SFN) functionality in 5G Networks is proposed. SFN transmissions impose several challenges and delay constraints in distribution networks. To overcome this, a new functionality is developed by inserting a new logical entity, Centralized Coordination Entity (CCE), inside Cloud-RAN architecture. By leveraging 3GPP Release 15 gNB split functionality, and extending eMBMS SYNC protocol this proposal has minimal imprint over 5G Networks and can support a wide range of SFN transmissions. This design fulfils the requirements for Vehicular and Mission Critical communications, Linear TV distribution or IoT, while overcoming existing limitations in Release 14 eMBMS. A complexity analysis of this solution in terms of imprint over 5G Architecture, latency and dimensionality is used to validate the proposal.

Rapidly Deployable Network System for Critical Communications in Remote Locations


Most of the authorities, especially within public safety, currently use dedicated radio networks for critical communications. Typically public safety communication services are narrowband professional mobile radio (PMR) systems, such as terrestrial trunked radio (TETRA) in Europe. The demand within the public safety is however shifting towards broadband services, such as video and multimedia, as they can provide useful information in many critical scenarios. As the dedicated narrowband PMR systems fail to meet this demand, the trend is going towards commercial mobile broadband networks, such as 3GPP long term evolution (LTE) and 5G technologies [1]. A major advantage involved in using commercial mobile broadband networks is the economies of scale achievable with them when compared to existing public safety communication technologies [2]. Commercial mobile radio networks are a cost-effective option for critical communications due to a large ecosystem and a wide selection of affordable user equipment (UE).

The network coverage is a critical requirement for critical communication networks. In remote locations, where the commercial network may not be available at all or is temporarily unavailable, transportable and rapidly deployable networks (RDNs) could be used to provide the required network coverage. However, providing a backhaul connection to the Internet can be a challenge for RDNs.

The trials described in this paper demonstrate a rapidly deployable LTE-based network built using only commercially available devices and software. Our RDN is designed to provide local voice, video and data connectivity for critical communications and Internet connectivity through commercial mobile networks (when available). The main features of the system are the following:

• Transportable. • Rapidly deployable; typically 15 to 45 minutes depending on the system complexity. • Stand-alone; independent of electricity and Internet. • Built with commercial off-the-shelf equipment and software.

The terminals connected to the network can communicate between each other (audio, video streaming, etc.) even without a backhaul connection, as the used evolved packet cores (EPCs) of the RDN do not require a connection to the Internet. The trials were conducted in Turku archipelago in Finland, as shown in Figure 1. Link distances with up to 15 km sealand propagation path were trialed using a towable person lift as an antenna mast with a maximum height of 27 meters above sea level, as seen in Figure 2. Previous tests show that LTE in maritime environments can achieve a link distance of around 100 km [3].

The trials consisted of radio signal measurements on 700 MHz and 2300 MHz bands and application testing of commercially developed solutions: Bittium Tactical Voice Service, Airbus Push to Talk application [4] and Fairspectrum Licensed Shared Access (LSA) evolution spectrum manager. LSA evolution is a framework which is currently being developed to provide user priorities and more dynamic overall approach to access shared spectrum resources than the somewhat limited and static LSA framework developed for 2300 MHz band can provide [5]–[7]. The trialed LSA evolution spectrum manager can control and dynamically manage the priorities of the transmitters connected to it. Thus the trials simulated a scenario where the spectrum for critical communication transmissions is prioritized over the other transmissions within the band. This means that the lower priority transmissions can be moved to other parts of the spectrum or completely shut down when the higher priority critical communications need the spectrum resources of the lower priority transmissions.


A high level illustration of the trial system is shown in Figure 3. A base station operating at the 700 MHz band was installed on the Kustavi coast in the upper right corner of the figure to study the long-distance LTE-link reliability and quality of service over sea or mixed land-sea paths in coastal and archipelago environment. The 700 MHz base station was also connected to the commercial Internet to provide Internet connectivity to a ferry and a smaller measurement vessel at the sea, where the 700 MHz signal was received to study the RDN coverage and data rate. A 700 MHz 4G mobile router was used to route the Internet traffic to a 2300 MHz base station on the Inio island in the ¨ lower left corner. The 2300 MHz base station was dedicated to provide local connectivity to the LTE UEs in the Inio island ¨ and its vicinity.

A. The trial system A block diagram of the trial system is shown in Figure 4. The diagram and the used equipment will be described in detail in the full version of the paper. III. MEASUREMENT RESULTS Conducted radio signal measurements were: • Transferring audio and high definition video over long distances with the 700 MHz link. • Data rate and connection quality as a function of distance. • Radio coverage over sea and land-sea paths. The measurement results will be presented and discussed in the full version of the paper.


This section briefly discusses potential improvements for a 3GPP-based RDN which were outside the scope of the trials presented in this paper, but could be essential for critical communications networks. A wider overview on the history and future of critical communications in 3GPP networks is provided in [1]. 5G makes it possible to prioritize the critical communication transmissions within a commercial mobile network through network slicing and quality-of-service (QoS) control [1]. The trials of this paper considered the priority of critical communications only with respect to the spectrum used for the transmissions. The 3GPP technologies also include a broadcast capability, which will be essential to critical communication networks when they are streaming situational awareness information to a large amount of terminals. For example, high definition video transmissions through unicast could congest the network [8]. The broadcast capability in LTE, called evolved Multimedia Broadcast Multicast Service (eMBMS), is somewhat static and limited [9]. Thus, a more flexible point-to-multipoint (PTM) framework, which integrates point-to-point and PTM modes under one common framework and enables dynamic use of PTM to maximize network and spectrum efficiency, is under development [10]. This new PTM framework will be important in avoiding network congestion when broadcasting situational awareness information in future 5G critical communication networks. Spectrum sensing technologies could also be used to validate the LSA spectrum information and to select the best channel if the LSA system is not available.

V. CONCLUSIONS The trials described in this paper indicate that it is possible to build a transportable, rapidly deployable and standalone LTE Network using commercially available hardware and software products. The network is independent of electricity and Internet and can be built in 15 to 45 minutes, depending on the required system complexity. Further trials are planned to investigate the coverage and QoS with for example different IoT technologies, such as NB-IoT and LTE-M, and 5G New Radio.

10:50-12:30 Session 7B: Resource Allocation and Access

Resource Allocation and Access

A Distributed Energy-Efficient Resource Allocation Mechanism for Multimedia Broadband Services in 5G Ultra-dense Networks

ABSTRACT. Multimedia Broadband Services (MBS) in 5G introduced by new smart devices will lead to the explosive mobile traffic growth. As one of the 5G new radio concepts, Ultra-dense Networks (UDNs), in which Small-cell Base Stations (SBSs) are densely deployed throughout the Macro-cell Base Stations (Macro-BSs) covered area, can improve the overall network capacity and has been recognized as one potential solution for MBS. However, the densification of large numbers of Base Stations (BSs) faces huge energy consumption challenge. Therefore in this paper, we propose a distributed energy-efficient resource allocation mechanism for 5G UDN meeting MBS users’ QoS. First, we use K-means++ algorithm to cluster the BSs in the network into several clusters, reducing the complexity of resource allocation range. Then, we propose a BS-sleeping algorithm to turn the BSs with lower traffic into sleep mode in each cluster, reducing the overall network energy consumption. Finally, we adopt an iterative transmit power optimization method to obtain the optimal transmit power of BSs in each cluster, further improving the Energy Efficiency (EE) of UDN by maximizing the EE of each cluster in a distributed way. Simulation results show that the proposed approach can effectively reduce the energy consumption and can improve the EE of UDN by nearly 23% while guaranteeing the QoS of MBS User Equipments (UEs).

A Novel Random Access Mechanism based on Real-time Access Intensity Detection

ABSTRACT. The Random Access (RA) procedure in 4G LTE serves for achieving uplink synchronization between UE and eNB and allocating channel resource for data transmission. The existing 4G LTE RA Channel (RACH) lacks the adjustment to the real-time RA traffic and suffers from preamble collision and system congestion caused by the RA request flooding, which will not meet the requirement of ubiquitous and massive connection in 5G. To better improve the system throughput and the RA success probability of RACH in the dense device network, we propose a novel congestion-aware RA mechanism in which concurrent devices carry out a Real-time Access Intensity Detection (RAID) prior to RA preamble message via a two-phase process. In this paper, we develop an analytical model based on stochastic geometry and derive the RA success probability of our proposed model for typical device in single RA slot. The analytical results demonstrate the improvement of our proposed RA mechanism in terms of RA success probability under heavy RA traffic. Furthermore, a large amount of simulations under homogeneous and cluster Poisson Point Process (PPP) is carried out to analyze the performance of the proposed RA mechanism in terms of the system throughput, the RA success probability, the retransmission time and the RA delay.

An Efficient Resource Allocation Mechanism for Beam-hopping Based LEO Satellite Communication System

ABSTRACT. LEO satellite communication network has become a promising approach to provide broadband access to internet service in low infrastructure areas. However, the LEO satellites are generally cost-constrained and the relative motion between the earth and LEO satellite is high-speed. As a result, the LEO satellite faces limited on-board resources and changing traffic demand distribution. In this paper, we design a beam-hopping system for LEO satellite communication, where the onboard resources are allocated flexibly according to the traffic demand distribution, which can improve the performance of throughput and onboard resource utilization. Specifically, a beam-hopping based LEO satellite communication system is constructed and the resource allocation problem is formulated. Then, we propose a greedy algorithm to resolve it. Evaluation results show that the beam-hopping system outperforms the fixed beam system and random beam hopping system in terms of throughput and request satisfied ratio.

Joint Caching Allocation and Delay Control for Network Slices Carrying Multimedia Services in 5G Core Network

ABSTRACT. As one of the key technologies of 5G, network slicing has received extensive attention and research in recent years. Infrastructure providers (InPs) can divide the physical network into multiple network slices, and service providers (SPs) can rent the network slices from InP to meet their own requirements. Since most multimedia services usually require low latency, such as virtual reality service, mobile high-definition video service and so on, caching resource can be deployed to network slicing to reduce the transmission delay and improve the quality of experience (QoE) of end-users. Due to the amount of caching resource is usually limited, how to efficiently allocate caching resource to save capital expenditure (CapEx) and operating expenses (OpEx) of InPs is an important issue. In this paper, we jointly consider network slicing and caching resource allocation, and propose a caching resource allocation scheme based on Benders decomposition algorithm. It can maximize the revenue of InPs while satisfying the requirements for caching resource and delay of each network slice. Finally, extensive simulation results are presented to illustrate the performance of the proposed scheme.

10:50-12:30 Session 7C: HEVC, VVC and Video Coding

HEVC, VVC and Video Coding

Enhancement on the Entropy Coding of Intra Mode in VVC

ABSTRACT. The emerging Versatile Video Coding (VVC) standard introduces several intra prediction modes and intra mode coding techniques. Most Probable Mode (MPM) is derived to enable efficient signaling for the selected intra mode. We propose the modification of intra mode coding for the MPM index. The first bin of mpm_idx is adaptively coded by regular coding engine or bypass engine according to neighboring intra modes. Two versions of the VVC Test Model (VTM) 2.0 and VTM4.0 were used to verify the proposed method with different numbers of MPM. Experimental results in 3-MPMs show 0.10%, 0.11%, 0.12% BD-rate savings for Y, U, and V components, respectively, for All Intra (AI) configuration. And the other results in 6-MPM show 0.05%, 0.08%, 0.07% BD-rate gains. No noticeable changes in runtimes of both encoder and decoder are observed.

Temporal dependency based CTU-level bit allocation for rate control

ABSTRACT. Rate control ensures the coded bit stream can meet channel bandwidth or storage space requirements. however, it may degrade rate-distortion (R-D) performance of encoders. Inspired by temporal dependent rate distortion optimization, this paper proposes a temporal dependency based CTU level rate control algorithm to improve R-D performance while achieving a high bitrate accuracy. First, an estimation method of CTU level R-D model is proposed to better characterize R-D relationship. Second, a formulation combining inter-CTU dependencies and R-D characteristics is constructed for CTU level bit allocation, which aims to minimize the distortion of the whole video sequence under the constraint of frame level bit budget. Finally, in the process of encoding a frame, a strict constraint of bits is used to reduce the difference between numbers of actual and target bits for the current frame. Experimental results show that, on the platform of HEVC test model (HM-16.7), the proposed algorithm achieves considerable R-D performance improvement with smaller bitrate error. Specifically, against HM-16.7 without rate control, the proposed scheme achieves bitrate savings of 4.2% and 3.7% on average under the configurations of low delay P and B frames, respectively.

Simplified Level Decision in Rate-distortion Optimized Quantization

ABSTRACT. This paper addresses a simplified rate-distortion optimized quantization scheme for Hight Efficiency Video Coding (HEVC) encoder. A level estimation (LE) bypass scheme and early termination of the last nonzero coefficient search are discussed to save the number of rate-distortion cost calculation. The proposed LE bypass schemes achieve 0.03% BDBR gain together with 3.60% quantization time saving on average.

Effective Video Compression with two Quantization Parameters

ABSTRACT. In this paper, we propose the novel methodology to improve the rate-distortion characteristics by the difference of the quantized DCT coefficients using two different quantized parameters in HEVC. Under the special condition of the quantization property in HEVC, we evaluate the binary difference of the quantized DCT coefficients derived from two different quantized parameters and compressed it with simple CABAC algorithm based on the characteristics of the binary difference.

The experimental result shows that the proposed algorithm improves the compression performance of the rate-distortion property in comparison to the conventional HEVC compression method.

13:50-15:55 Session 8A: ATSC 3.0

ATSC 3.0

ATSC 3.0 Channel Bonding Performance in Mobile Channel Environments

ABSTRACT. This paper presents channel bonding performance of the ATSC 3.0 system in mobile channel environments. Channel bonding can be used to increase data rate of broadcasting services by utilizing additional RF channels. Moreover, it can also provide diversity gains by combining non-contiguous RF channels with different carrier frequencies. Simulation results show that system performance is significantly improved by employing channel bonding across different carrier frequency bands.

Development of MMT Analyzer for ATSC 3.0

ABSTRACT. With the increasing volume of video streaming on the Internet, whether through on-demand or real-time services, the need arose for the development of transport mechanisms and protocols with greater flexibility and control of the quality of the transmitted service. To meet this new issue and ensure the evolution of broadcast television services, MMT (MPEG Media Transport), was developed to transport the hybrid delivery of multimedia streams in broadcast and broadband networks. This paper proposes the implementation of a specialized MMT Protocol (MMTP) analyzer for ATSC 3.0 developed in C language and integrated as a plug-in for the Open-Source Wireshark software.

Detection Performance of ATSC 3.0 TxID in Multiple-Transmitter Environments

ABSTRACT. This paper provides analysis and test results on detection performance of Transmitter Identification (TxID) in Advanced Television Systems Committee (ATSC) 3.0 system. In the next generation of the terrestrial broadcasting system based on ATSC 3.0, it is likely that single frequency network (SFN) is widely deployed. In SFN environment, the detection performance of TxID can be degraded depending on the number of transmitters and its relative reception power from multiple transmitters. In this paper, in terms of TxID detection performance, the impact of multiple transmission signals is analyzed through the computer simulation, laboratory test, and field test.

A New Decoding Scheme of Emergency Alert Wake-up Bits in ATSC 3.0

ABSTRACT. This paper presents a new decoding scheme of emergency alert (EA) wake-up bits and evaluates its decoding performance in Advanced Television Systems Committee (ATSC) 3.0. Given that the minimal essential information in the bootstrap signal of ATSC 3.0 is not being changed often, the received signal can be further assumed to be binary modulated according to each wake-up bit. Therefore, the wake-up bits can be obtained independently regardless of other signaling information after the initial decoding that obtains entire information carried by the bootstrap. The simulation results show that the proposed decoding scheme of wake-up bits outperforms the conventional scheme which decodes the entire information carried by the bootstrap.

Improved Min-Sum Algorithm for Raptor-Like LDPC codes in ATSC 3.0

ABSTRACT. In order to improve the performance of low-rate LDPC codes, ATSC 3.0 adopted raptor-like structure. Min-sum (MS) algorithm and improved versions can be used directly, but their performance on raptor-like LDPC codes are disappointing. In this paper, we introduce an improved min-sum(IMS) algorithm for raptor-like LDPC codes in ATSC 3.0. Simulation results of frame error rate show that the IMS algorithm not only has a noticeable gain in waterfall region over normalized min-sum (NMS) algorithm and offset min-sun (OMS) algorithm, but also has almost the same time complexity. Different from OMS algorithm, the proposed algorithm is not sensitive to the imperfect channel estimation. Meanwhile, the IMS algorithm is easier to realize than those variable factors based algorithms.

13:50-15:55 Session 8B: New Media Services

New Media Services

Delivery of High Definition Video Content over Bandwidth Constrained Links in Heterogeneous Wireless Networks

ABSTRACT. I. INTRODUCTION Successful delivery of High definition (HD) video content to mobile users in heterogeneous, multi-network wireless environment is challenging since, by its very nature HD video requires high capacity reliable communication links. However, communications in wireless networks are characterized by unreliable communication links and rapidly changing levels of available bandwidth. End-user satisfaction with a video streaming service is closely tied to their quality of experience (QoE) which differs from the Quality of Service (QoS) provided by the networks and their operators.

QoE for the end-user is influenced by many factors including the initial delay in playing the selected video, the number and duration of any interruptions to the video playback (stalling events), the location of any stalling events within the video itself (start, middle or end of video), drastic shifts in the quality of the video, etc. The amount of bandwidth available at any one time and fluctuations in the bandwidth have a direct, powerful effect on the QoE of the viewer. In particular, consumers of HD video will experience a dramatic reduction in QoE if the content being viewed suddenly switches from HD to a low bitrate representation due to a reduction in available bandwidth.

Dynamic Adaptive Streaming over HTTP (DASH) is an example of an adaptive streaming strategy that has been implemented to address the challenges posed by large fluctuations in available bandwidth. DASH operates by segmenting the content to be delivered into chunks of various durations e.g. 2 seconds, 4 seconds, 6 seconds, 10 seconds and encoding each chunk at various bit rates. This process results in multiple copies of the content being stored on the HTTP server.

DASH enabled clients deal with fluctuations in the available bandwidth by dynamically adapting the bitrate of the content being downloaded to match as closely as possible the available bandwidth of the wireless link. In response to a reduction in the available bandwidth there is a reduction in the bitrate of the video segments being requested. However, in the case of HD video content there is a lower bound to the bitrate video segments below which the DASH client cannot go if the content is to remain HD.

The challenge research and development faces is designing solutions to maintain QoE for the end-user employing a DASH enabled client application when the lowest acceptable bitrate for HD video is greater than the capacity of the available bandwidth on a link. II. PROPOSED SOLUTION This paper proposes Constrained Bandwidth Video Delivery (CBVD) which leverages both the multi-homed nature of modern portable devices and the segmented content held on DASH servers to address the challenge. Modern smart-phones and other portable devices such as tablets are equipped with multiple wireless interfaces by default, these typically consist of a WiFi interface and a SIM card for connecting to cellular networks. Users of cellular data plans are constrained by monthly download limits known as data-caps, breaching these data-caps can result in excessive charges per MB of data over the limit and possible throttling of the connection. In 5G networks data-caps will prove to be even more of an issue for end users [1] due to service provider policies. For example, AT&T propose a mere 15GB data-cap for mobile users on their 5G networks in the US which due to the high data rates available will be exhausted in a short period of time particularly when streaming HD video content.

Under normal usage conditions the various interfaces of a device are used in isolation (Figure 1) but using both interfaces simultaneously would enable the user to increase the total amount of bandwidth available.

Figure 1: DASH segment transfer over single link

Instead of adjusting the bitrate of the content to match the available bandwidth we increase the available bandwidth to match the bitrate of the content. This is achieved by activating both interfaces simultaneously and load balancing the segment transfers over both links (Figure 2). Load balancing is a technique for transferring data across multiple resources such as physical or wireless interfaces.

Figure 2: DASH segment transfer over multiple links

The aim of load balancing is to maximise throughput and to reduce overloading any single link or interface. Load balancing can also increase reliability and availability through redundancy by using multiple components instead of a single component. For mobile users load balancing may come at an increased cost in terms of energy consumption and under certain circumstances can lead to an overall reduction in the total amount of data transferred. Previous work introduced the SONS [2] framework for determining the appropriate under which a mobile user should attempt to connect to a Wi-Fi network and when it should be avoided.

The segmented nature of the content on a DASH server in conjunction with the multi-homed capabilities of modern mobile device provide the basis of a solution to the problem. Typically, mobile users of smart devices will connect to WiFi where available in order to reduce the amount of data they download over their cellular connection. They do so to take advantage of often faster connection speeds, reduce their energy consumption and protect their data cap. In order to be able to request segments having a bitrate that can be accommodated on the prevailing wireless link the DASH enabled client must determine the throughput of the link. Monitoring the level of the playout buffer as demonstrated by [3] provides a simple means of inferring the capacity of the link without having to calculate the actual throughput.

If the level of the playout buffer falls too low, then the video playback begins to experience stalling events which are detrimental to the QoE of the viewer. Stalling events indicate that the link capacity is insufficient to support the current segment bitrate. In this case the system re-activates the cellular link and begins downloading the next segment in the sequence as indicated by the MPD or manifest. The playout buffer levels are rapidly increased by having alternating segments downloaded over each link until the playout buffer level has stabilized. When the playout buffer has regained a level at which the risk of stalling events occurring has been reduced to near zero the cellular link is de-activated and the buffer continues to be monitored. A new temporary lower trigger level is implemented so that the standby interface can be activated sooner if necessary when the buffer level drops in the future.

Figure 3: Virtualbox, NS3 and Linux Container Test environment

The test environment was deployed on a Lenovo ThinkPad equipped with an Intel i7 processor and 16GB of RAM running the Debian 9 Linux [4] distribution. An Oracle VM Virtualbox [5] Debian virtual machine (VM) had a modified version of the DASH Industry Federation JavaScript Reference Player [6] installed. The DASH enabled player requested content from the Apache server on a Linux Container (LXC) [7]. An NS3 simulated wireless network provided connectivity between the DASH player and the server (Figure 3).

The proposed solution will be evaluated against the standard single link approach in terms of the number of stalling events and the total amount of data transferred over the cellular connection. The total amount of energy consumed in both scenarios during content delivery will also be evaluated against the total number of stalling events. Fluctuations in the available bandwidth on the NS3 links will introduced to mimic real world conditions for mobile users.

CBVD maintains the end-users QoE by enabling HD video streaming in low bandwidth conditions while protecting the user’s data-cap and reducing energy consumption through intelligent network interface management.


[1], accessed December 20th 2018

[2] Casey, T., Muntean, G-M., “Scan-Or-Not-To-Scan – Balancing Network Selection Accuracy and Energy Consumption”, Proceedings of the 11th International Wireless Communications and Mobile Computing Conference, Dubrovnik, Croatia, August 24 – August 28, 2015

[3] Huang, Weiwei & Zhou, Yipeng & Xie, Xueyan & Wu, Di & Chen, Min & Ngai, Edith. (2018). Buffer State is Enough: Simplifying the Design of QoE-Aware HTTP Adaptive Video Streaming. IEEE Transactions on Broadcasting. PP. 1-12. 10.1109/TBC.2018.2789580.

[4] Debian,, accessed November 2018

[5] Oracle VM Virtualbox,, accessed November 2018

[6] DASH IF Reference Player,, accessed November 2018

[7] Linux LXC,, accessed November 2018

On Hit Rate Improving and Energy Consumption Minimizing in Cache-Based Convergent Overlay Network on High-speed Train

ABSTRACT. Content caching and energy consumption to protract user device lifetime while bolstering the hit rate when mobility is beyond 300Kmph is challenging. In this paper, we consider a cache-based convergent overlay network comprising cellular base stations and terrestrial broadcast networks as effective means to deliver the services in high-speed train(HST). Most popular contents with high Zipf rank are pushed and cached using broadcast network in user devices and relay system on HST to offload the network traffic and bring user experience upfront leveraging energy. Presence of cache in relay system (RS), users can get the services without delay. However, if user's request for contents are served by cellular network from cloud, they face throughput bottleneck with delay and increased transmission time. This eventually leads increase in user device energy consumption. A popular contents caching scheme with constraints of location, limited storage size and limited power of user device is modeled as closed form expression to maximize users local hit rate with minimal power consumption. we propose an algorithm to cache the contents in user device to minimize total energy consumption of user devices. Moreover, user's content activity behavior is retained to minimize user device's energy by dynamic cache space algorithm. Simulation results justify that the proposed scheme can effectively improve cache hit rate and reduce the power consumption of user devices.

Trials of 60 GHz Band for a 5G New Radio (NR) Solution for High Capacity CCTV Offload and Multimedia Transfer

ABSTRACT. This paper studies the performance of a 60 GHz link in both indoor and outdoor conditions. The target is to assess its suitability for resolving emerging needs in the public transport, especially, in rail traffic, to transfer large amounts of data from vehicles to the stations and vice versa, during a short period of time. 60 GHz could also be ideal band for the wireless inter-carriage connection between the railcars. The related services and requirements for the needs described in the latter are defined in the 5G specification Mobile Communication System for Railways - TS22.289. 60 GHz band is also included in the 5G standard as an unlicensed band.

Generation of Automated Sports Commentary from Live Sports Data

ABSTRACT. Live data on a wide range of sports competitions have recently been created and are currently provided as text, graphs, or synthesized voices conveying simple phrases via TV telops, datacasting, the Internet, etc. to help people enjoy sporting events. We have developed method of generating automated sports commentary that conveys a situation with a series of synthesized speech comments that are coherent throughout the game based on live data. Our proposed method of generating automated sports commentary continually conveys the differences between the latest data and past-conveyed facts vocally in order for listeners to follow the current situation of the game quickly. We conducted a subjective evaluation to confirm its usefulness., which indicates that automated sports commentary generated with our method is as helpful in understanding a game.

MEC-assisted Immersive Services: Orchestration Framework and Protocol

ABSTRACT. Immersive media applications demand both a huge amount of data processing and low latency. The former requirement can be fulfilled by offloading user equipment (UE) tasks to a baseband unit (BBU) pool (or data center). As for the low latency, it can be achieved with Mobile Edge Computing (MEC), which allows offloading the tasks close to the UEs. In this paper, we take advantage of the cooperation between conventional BBUs and MEC to handle the UE tasks. Specifically, we propose an orchestration framework and a protocol to enable seamless immersive media applications. Our extensive simulations show that our solution can significantly reduce the number of failed tasks.

13:50-15:55 Session 8C: Multi-View and Image Processing

Multi-View and Image Processing

Ice Hockey Player Tracking and Identification System using Multi-view Video

ABSTRACT. 1. INTRODUCTION During ice hockey game, there is frequent position overlapping between players due to body-check. This causes position errors of ice hockey players when we track them with single camera. In order to alleviate this problem, we propose a ice hockey player tracking and identification system that utilizes a multi-view video and a jersey number identification technique based on color model and SVM (support vector machine). 2. PROPOSED ICE HOCKEY PLAYER TRACKING AND IDENTIFICATION SYSTEM The proposed system utilizes at least 4 camera input video for robust ice hockey player tracking. Ice hockey player tracking process is performed in a hierarchical manner and consists of two stages: local player tracking and global tracking. In the local tracking stage, ice hockey player extraction and tracking is performed in each camera image individually. In the global tracking stage, tracked positions of each player in each camera image are integrated and cross-checked across the camera views. Finally, only validated positions of each player are recorded to database and used for motion analysis of each player. Even if we use multi-camera video-based tracking, some players could be missed or be switched incorrectly to another player. For improving this kind of tracking errors, we investigate the feasibility of player identification technique through jersey number recognition. We adopt the RGB color model for jersey number region extraction, and SVM(support vector machine) technique for jesery number recognition. Training and test sample images are captured from the tracking cameras or a separate camera. After updating player’s positions with jersery number recognition, we store each player’s position data and use them for the analysis of each player’s performance statistics such as average speed, averge accerlation. These data can be visualized with graphs, heat map, and trajectories. In addition to motion anlaysis of each ice hockey player, proposed system provides additional functionalities. For example, player’s region tracking-based camera calibration, a event tagging, event video segment extraction, and ROI(Region of Interest) view selction. These functionalites are also useful in analyzing a ice hockey game and each player’s performance during the game according to the survey on the national ice hockey team coaches. Player’s region tracking-based camera calibration uses a extracted center line of each player obtained from player extraction and tracking information for the camera calibration of each camera. By calcuting a homograpy matrix beteween camera view images with extracted center lines, we obtain the camera parameters without camera calibration pattern. Event tagging and event video segment extraction functions provide the extraction of important game events during the game for the game analysis. Start time and end time of each tagged event and a correponding video segment clip are stored in the database and it can be replayed at any time. ROI view selection function helps to make a highlight video of the game by selecting a ROI view. Three cameras are used to acquire ROI view candidates in the ice rink. The field of view of each camera covers left, center, and right part of ice rink with partial overlaps, respectively. During the game, the proposed system tracks ice hockey players and counts the number of players in the each ROI candidate view image. And then it selects one of ROI view candidates that has maximum number of players among left, center, and right ROI candidate views. Optionally, we can also select one ROI view manually. 3. EXPERIMENTAL TEST RESULTS We implemented and installed it at the ice rink of the Jincheon National Training Center. 8 cameras for ice hockey player tracking are installed at near the ceiling of ice rink, and 3 cameras for game event tagging, event segment extraction, and ROI view selection are installed at the upper side wall. Another one camera for player jersy number recognition is installed on the balcony of rocker room located at 4th floor of the ice rink. The ice hockey player tracking and motion analysis server is also installed in the rocker room. For the syncronization of 8 camera for tracking, an external 5V trigger signal is applied to each camera. Each camera video is acquired with a video acquisition software and two 4-port GigE frame grabber boards. Acquired videos are encoded with MPEG-4 AVC and stored with mp4. The maximum resolution of each input video is 1,920 x 1,080, and its maximum frame rate is 30fps. For the faster ice hockey player tracking, a player tracking software firstly reduces the size and the frame rate of the input image in half. In addition, multi-thread is used to assign a player tracking module for each camera view. As described in section 2, each ice hockey player’s position in each frame is regarded as vaild when tracked player’s positions extracted from more than 4 cameras almost agree with each other. We firstly measured the average tracking speed in fps when we set the number of cameras used for player tracking with 4, 6, and 8. As a result, the average tracking speed was 16.6, 10.4 and 9.0 fps, respectively. As a next experiment, we tested the proposed player jersey number extraction and recognition technique with 251 test jersey number images. In 251 test images, some images are blurred or rotated or distorted. Even if there exists such image distortions, 248 jersey number images were recognized correctly and the average jersey number recognition rate was 98.8 %. 4. CONCLUSION We presented a ice hockey player tracking and identification system using multi-view video. The propsed system provides useful functionalities for ice hockey game analysis. And It showed good performance in terms of tracking speed and jersey number recognition rate. The propsed system can be applied to an additional multimedia service in sports game. ACKNOWLEDGEMENT This work was supported by ICT R&D program of MSIP/IITP. [2017-0-00543, Development of Precise Positioning Technology for the Enhancement of Pedestrian’s Position/Spatial Cognition and Sports Competition Analysis].

Color and Illumination Compensation Scheme for Multi-view Video Service
PRESENTER: Jongbin Park

ABSTRACT. This paper deals with the problem of color and illumination compensation for multi-view video service. In particular, we focus on finding a simple and lightweight color compensation algorithm because it is more friendly for real-time service. In this paper, we developed multi-view rendering software and applied and tested the diagonal matrix based color compensation method. Through experiments, we confirmed that there are advantages of diagonal color transfer model compared with histogram matching and statistical color transfer scheme. We also have experimentally observed that the approach works even when the multi-view image is corrupted under realistic conditions such as rain or sun-flare.

Digital Meter Recognition Method Based on Topological Features of Image Contour

ABSTRACT. With the rapid development of production automation and industrial intelligence, automatic recognition of digital meter readings has played a significant part in this trend of information revolution. To meet the higher and higher demand of recognition with efficiency and accuracy, a method based on topological features of image contour is proposed. Image contours are classified according to size and bounding hierarchy during the reading location process, from which expected digits can be selected with the priori knowledge of the meter. The effectiveness of the algorithm is experimentally validated in complex environments.

Low-precision CNN Model Quantization based on Optimal Scaling Factor Estimation

ABSTRACT. With the development of convolutional neural net-works (CNNs), researchers have acquired satisfactory performances on computer vision tasks such as image classification and semantic segmentation. However, CNNs contain millons of weight parameters and demand huge storage and computation resources, making it hard to deploy on constrained hardwares. To compress the model and accelerate the deploying speed, many model compression methods have been proposed, including quantization methods that aim to reduce network redundancy by decreasingthe representational bits for weights (and activations). However, for low-bit quantization, the inevitable quantization error will lead to significant accuracy degradation and gradient mismatch. In this paper, we propose Scale Estimation Quantization (SEQ). To reduce the quantization error, we analyze the variance of errorderived by the quantization process. By exploiting the distributions of network values, we minimize the quantization error and estimate the optimal scale parameters for our proposed quantization function. Further more, to deal with gradient mismatch in backward propagation, we propose backward approximation. We apply our algorithm on image classification tasks. Without any accuracy degradation, our method successfully compresses VGG-Small With 1-bit weights and 2-bit activations.

Protecting Multimedia Privacy from Both Humans and AI

ABSTRACT. With the development of artificial intelligence (AI), multimedia privacy issues have become more challenging than ever. AI-assisted malicious entities can steal private information from multimedia data more easily than humans. Traditional multimedia privacy protection only considers the situation when humans are the adversaries, therefore they are ineffective against AI-assisted attackers. In this paper, we develop a new framework and new algorithms that can protect image privacy from both humans and AI. We combine the idea of adversarial image perturbation which is effective against AI and the obfuscation technique for human adversaries. Experiments show that our proposed methods work well for all types of attackers.

16:15-17:30 Session 9: POSTER


Focal Stack based Light Field Coding for Refocusing Applications

ABSTRACT. Since light field data captured by the plenoptic camera has a huge data amount, it requires an efficient compression method for transmission and storage. In refocusing applications, users only wants to see their interested object at different depth levels. Therefore, we first render a set of refocused images at all focal points, then compress them instead of original light field data. By doing so, we can minimize the amount of data to process in target refocusing applications. Our experiment results show that our proposed method can save 47.6% of bitrate and gain 3.87 dB in term of PSNR at the same bitrate

Performance Evaluation of AV1 Intra Coding Tools

ABSTRACT. To meet the growing requirement of servicing high-quality video contents, Alliance for Open Media (AOMedia) has recently released an open-source video coding scheme named as AOMedia Video Coding 1 (AV1). In this paper, we review its intra coding tools in details and evaluate their individual performance in On-Off tests. Our experimental results show that the additional directional mode and the recursive filtering achieve the highest Y-BDBR gain under all intra and random access configurations, respectively. The chroma from luma prediction saves significant bits in chroma channels. Additionally, the Paeth predictor performs betters for screen contents. Color palette prediction is seen effective for screen content sequence.

Novel Method for broadcasting Multiples Emergency Alert and Alarm Using ISDB-T Digital Terrestrial Television

ABSTRACT. In the Digital Terrestrial Television System ISDB-T (Integrated Services Digital Broadcasting Terrestrial), it is specified how to send emergency information by the Emergency Descriptor EWBS (Emergency Warning Broadcast System) of Table PMT (Program Map Table); The emergency information is restricted to 1 bit with status of (activation/No -activation) and the area code that determines the geographic location where the alert or alarm will be activated.. The present paper proposes a novel method to send information of alerts and emergency alarms contemplating multiple events and complementary information, this is achieved by proposing a data protocol that will be responsible for bringing complementary information to the existing one. A new table is defined within the Transport Stream MPEG with a data structure that allows sending this additional information to indicate new types of emergencies.

Implementation and Performance Evaluation of a MIMO-VLC System for Data Transmissions

ABSTRACT. The ever-increasing streaming culture of large amounts of data and the need for faster and reliable methods of data transfer has created a space and market for new communication technologies such as Visible Light Communication (VLC). However, the integration of VLC into next generation networks is challenging due to the drawbacks of the technology in terms of atmospheric absorption, shadowing, beam dispersion, etc. One way to overcome some of the challenges is to make use of the multiple input multiple output (MIMO) technique which involves the transmission of data in parallel from multiple sources, increasing the data rate. This paper implements and provides a comprehensive evaluation of a MIMO-VLC system for data transmission. A real experimental test-bed is setup to test the performance of the MIMO-VLC system under various conditions such as distance from the source based on luminous flux, ambient lighting, output power, etc. Additionally, subjective tests are carried out to assess the quality of an audio MIMO-VLC link as perceived by the user. The results are also compared with the results from a Single Input Single Output (SISO)-VLC system.

PAPR reduction based on deep autoencoder for DCO-OFDM VLC system

ABSTRACT. DC-biased optical orthogonal frequency division multiplexing (DCO-OFDM) is widely used in visible light communication (VLC) systems to reduce the effects of inter-symbol interference (ISI) and to achieve high speed data transmission. However, DCO-OFDM VLC systems suffer from peak-to-average power ratios (PAPRs) issue which causes serious degradation of system performance. In this paper, a novel deep autoencoder-based PAPR reduction scheme for DCO-OFDM VLC is proposed. The autoencoder could adjust adaptively the constellation mapping and demmaping in order to jointly minimize the PAPR as well as BER for VLC system. The proposed scheme outperforms conventional methods in terms of BER and PAPR.

Positioning System based on DTMB: Design, Laboratory Test and Field Trail

ABSTRACT. Compared with Global Navigation Satellite System (GNSS) which has been deeply studied and widely applied, the systems based on digital terrestrial television broadcasting (DTTB) for positioning need to be further studied. A positioning system based on digital television terrestrial multimedia broadcasting(DTMB) system is well designed and implemented on FPGA in this paper. The designed postioning system is based on time difference of arrival (TDOA) measurements. Laboratory trail results show that the proposed system can provide accurate positioning results when the signal-to-noise ratio (SNR) is above 8 dB. Field trial results prove that the positioning error is less than 10 meters under line-of-sight (LOS) channels.

A Frame Loss Concealment Solution for Spatial Scalable HEVC using Base Layer Motion

ABSTRACT. Scalable High Efficiency Video Coding (SHVC), is the most recent video coding solution mainly designed for network adaptive or device adaptive applications. SHVC follows a layered coding structure with one base layer (BL) and one or several enhancement layers (ELs) which can be unequally protected. However, SHVC is often sensitive with the packet loss problem in unreliable networks, especially at the ELs. In this paper, we propose a novel error concealment method for the SHVC EL with an assumption that the BL is well protected. First, we resample the motion and residual data from collocated BL frame. Then, we remove outlier motion by a motion vector refinement algorithm. Lastly, the concealed frame is generated by using motion compensation and deblocking filter. Experiments conducted for a rich set of test sequences and for the most popular spatial – SHVC standard have shown that our proposed method significantly outperforms the relevant error concealment methods, e.g., BL Reconstruction Up-sampling (RU) and BL SKIP on both subjective and objective quality assessments.

A Practical Channel Estimation Scheme for Sub-Connected Hybrid Massive MIMO Systems

ABSTRACT. In massive MIMO system, the increasing number of antennas causes high power consumption and hardware complexity. Sub-connected hybrid system structure can effectively mitigate integration difficulty and reduce the number of radio frequency (RF) chains. In this paper, we proposes a low-overhead scheme for uplink channel estimation where a single-antenna user contacts with a sub-connected hybrid massive MIMO base station (BS). Since the sparsity caused by severe path loss, channel state information can be reconstructed by estimating the angles and gains of paths. We apply joint beam selection and sum-difference beam (JBSSDB) method to estimate path angles and utilize least square (LS) method to estimate gains information. Simulation results demonstrate that the proposed scheme achieves lower NMSE than conventional method in the case of same overhead.

New Semi-Blind Channel Estimation for LDM-LSI

ABSTRACT. Local content insertion and viewer segmentation with proximity service information is a critical business case for broadcasting operators. This personalized service delivery will open up an opportunity for new use cases, such as local service delivery, local advertisement insertion and local emergency warning. In practice, Single Frequency Networks (SFNs) offer a better spectrum efficiency when compared with traditional Multiple Frequency Network (MFN) approaches. Nevertheless, the addition of local content delivery can dramatically reduce that efficiency. Recently, the LDM based LDM-LSI solution has been proposed as an interesting alternative for including local content in single frequency networks. Nonetheless, the required channel estimation procedure has been considered a major drawback for its successful implementation. In this work, a new semi-blind channel estimation algorithm is proposed. The algorithm is based on the statistical properties of the time domain OFDM frame and solves the identified major challenges that have remain unsolved so far. The paper also includes the first simulation results for this new proposal.

Capacity Analysis for LDM-Based Multiple-PLP Configurations in ATSC 3.0

ABSTRACT. This paper analyzes capacities for various layered division multiplexing (LDM)-based multiple-physical layer pipe (M-PLP) configurations in Advanced Television Systems Committee 3.0. Also, this paper proposes transmitter and receiver structures for multi-layer LDM. The proposed multi-layer LDM consists of one core layer and one or more enhanced layer(s) that can be combined with different power levels. The simulation results show that LDM with an appropriate injection level can improve the capacity for M-PLP compared to TDM-based PLP configurations.

Implementation of a real-time ISDB-TB LDM receiver using SDR

ABSTRACT. This paper presents an implementation of a real-time ISDB-TB LDM receiver in Software Defined Radio/GNU Radio Companion. The main idea of this work is to provide a hybrid digital TV broadcasting system that can be used as an option in a transition period between the current Brazilian system (ISDB-TB) and a possible new standard, such as a modified version of ATSC 3.0. The Core Layer is fully compatible with ISDB-TB and is multiplexed with an Enhanced Layer using a powerful channel coder, high order modulations and some other ATSC 3.0 features and techniques. With the use of the LDM in such systems, it is possible to deploy UHDTV applications. The implementation is done using the GNU Radio Companion software along with a Software Defined Radio platform to transmit and receive the Radio Frequency signal in real-time.

DVB-T2 receiver: impact of FFT window positioning on radio coverage
PRESENTER: Subodh Nepal

ABSTRACT. In modern digital terrestrial broadcasting systems, Single Frequency Network (SFN) is a prominent technique used for improved spectrum efficiency, network gain, and increased flexibility. In an SFN, the placement and positioning of FFT window relative to the multiple received signals strongly affect the receiver’s behavior of handling Inter-Symbol Interference (ISI). Thus, the treatment of ISI in receiver further impacts on the radio coverage predictions and network planning. In this paper, the radio coverage due to a typical SFN network is evaluated and analyzed for different FFT window synchronization methods. This is repeated for different transmission and reception modes used in the next generation of broadcasting standards and systems specifically in the DVB-T2 system.

Multi-level Stereo Attention Model for Center Channel Extraction

ABSTRACT. In recent years, the spatial audio reproduction of digital media has become popular. Despite the demand for such spatial audio content, very little content is produced with multi-channel audio. Moreover, it is difficult to provide interactive services to users owing to the lack of object-based content. In this paper, we propose a center channel extraction method based on a multi-level convolutional neural network structure to generate object-based content. In addition, we present a novel stereo attention model which considers each channel’s characteristics. By applying the proposed method to stereo audio content, we achieve better extraction performance than existing commercial application.

NOMA-based 802.11n for Broadcasting Multimedia Content in Factory Automation Environments
PRESENTER: Eneko Iradier

ABSTRACT. Multimedia content distribution and, specifically, video consumption will be one of the most relevant factors for the traffic increase in wireless networks in the next following years. The environments where video is gaining momentum are not restricted to traditional video consumption and social networks. A new research area, which is rapidly developing and growing, is the field of industrial communications. Communications in industry have been traditionally oriented to control and alarm systems. Nevertheless, the range of communication use cases in industry is widening and video content will be needed sooner than expected. An innovative and original idea to combine the reliability requirements of the industry and the transmission of multimedia contents is to use NOMA over the physical layer of the 802.11n standard. This paper is the first comprehensive analysis of a video broadcasting/multicasting system within an industrial environment and based on an 802.11n-NOMA physical layer. To check the performance of the proposal for broadcasting multimedia delivery, an evaluation scenario is presented and simulations have been carried out.

Implementation of Targeted Advertisement Service on ATSC 3.0 Runtime Environment

ABSTRACT. This paper presents an application service platform for targeted advertisement in ATSC 3.0. The implemented platform is fully compatible with ATSC 3.0 standards and is comprised of three parts: a transport system, a receiving system and an application. The laboratory test results show that advertisement contents are successfully delivered in real-time and non-real-time over both broadcast and broadband networks, and personalized client-side advertisement switching is also performed based on application over ATSC 3.0 runtime environment.

Low-Complexity Multiuser Detection for Generalized Media-Based Modulation Systems

ABSTRACT. In this paper, we consider multiuser generalized media-based modulation (GMBM) system, where the users adopt both GSM and MBM schemes to convey extra information via antenna and RF mirror indices. The GSM scheme activates only a subset of available antennas for transmission, and the MBM scheme uses the on-off state of RF mirrors placed around the antennas to create different channel realizations. The combined transmission technique, termed GMBM, improves the spectral efficiency for transmission with limited RF resources. We propose two structured compressive-sensing algorithms to detect the multiuser signals with low complexity. The three types of structure, namely the users, the antennas, and the RF mirrors, are recovered during each iteration of the algorithms. The structured SP algorithm for MU-GMBM is shown to have better performance with more users, and both algorithms benefit from the increase in receive diversity.

Field trial of UHDTV over Digital Television Terrestrial Broadcasting network

ABSTRACT. The field trial of 4K UHDTV over DTMB-A system has been successfully conducted. Results show satisfactory performance for both indoor and outdoor reception environment with acceptable margins.

AROMA: An Adapt-or-Reroute Strategy for Multimedia Applications over SDN-based Wireless Environments

ABSTRACT. To support new and advanced multimedia-rich applications and services while providing satisfactory user experience, the underlying network infrastructure needs to evolve and adapt. One of the key enabling technologies of the next generation (5G) networks is the integration of Software Defined Networking (SDN) within a heterogeneous wireless environment to enable interoperability and QoS provisioning. Leveraging on the features of the SDN paradigm, it is possible to introduce new solutions to handle the increasing mobile video transmission challenges with strict QoS requirements, such as: low delay, jitter, packet loss, and high bandwidth demands. However, degradation and instability perceived from video traffic makes it difficult to satisfy various end-users. In this context, this paper proposes AROMA, an Adapt-or-reROute strategy for Multimedia Applications over SDN-based wireless environments. AROMA enables QoS provisioning over multimedia-oriented SDN-based WLAN environments. The proposed solution is evaluated using a real experimental test-bed setup.

A Study on Improving Extended Parity Generation for LDM-BST-OFDM

ABSTRACT. In this paper, we studied improvement of extended parity generation for LDM-BST-OFDM(Layered Division Multiplexing - Band Segmented Transmission – Orthogonal Frequency Division Multiplexing). In the proposed scheme for fixed reception, the parity-check matrix of the LDPC(Low-Density Parity-Check) code whose length is 69120 is extended. After that, the extended parity bits are generated by the extended parity-check matrix and modulated. These modulated symbols are LDM-combined with the symbols for mobile reception. The performances of the proposed scheme are evaluated by computer simulations under AWGN(Additive White Gaussian Noise). As the results, the required CNR(Carrier to Noise Ratio) of fixed reception can be improved by the proposed scheme.

3D Point Cloud Attribute Compression Based on Cylindrical Projection

ABSTRACT. With the rapid development of 3D acquisition technologies, there are mounting interests in representing real-world objects as 3D point clouds. However, nonuniform point positions and abundant photometry attributes of point cloud (i.e., colors, reflectance and normal directions) result in massive data volume, leading to the overload of transmission. To address this issue, we propose an attribute compression method for static 3D point cloud by taking advantage of 2D video coding technology. The attribute is projected on 2D images which are generated from cylindrical surface for several times. We require consistent angle for each projection to guarantee the spatial distribution of multiple projected images is similar, then we combine these images into video sequence. 2D video codec is utilized to perform compression. In this way, we benefit from both intra and inter prediction. Objective evaluation shows significant improvement in coding efficiency over reference technologies.

Gradient Bit-Flipping LDPC Decoder for ATSC 3.0

ABSTRACT. This paper presents the mathematical representation and the numerical results of the Gradient Bit-Flipping Algorithm (GBF) that is derived from the Bit-Flipping Algorithm (BFA). The experimental simulation results for the proposed decoder of the Low-Density Parity-Check (LDPC) codes of ATSC 3.0 demonstrate an enhancement of up to 3.55 dB in error correction performance without an increase in the decoding complexity. The proposed decoder has the ability to decode an erroneous LDPC message in fewer iterations than the traditional BFA due to the employment of the values of the Log-Likelihood Ratio (LLR) as an input, unlike the BFA that uses binary values ignoring the channel estimation output.

Analysis of Channel Capacity for Three-layer Signals in Layered Division Multiplexing System

ABSTRACT. A layered division multiplexing (LDM) system provides two or more services over one RF (radio frequency) channel. The LDM system draws much attention as a next generation broadcast communication system because it can provide 3 to 6 dB higher system gain than a time division multiplexing (TDM) or a frequency division multiplexing (FDM) systems. Many studies have been conducted to improve the transmission capacity and the performance of the two-layer LDM system. In order to achieve the additional transmission capacity, however, it is necessary to study terrestrial broadcasting system adopting LDM system with more than three layers. In this paper, we present the feasibility of a three-layer LDM system by analyzing its spectrum efficiency according to signal-to-noise ratios (SNRs) and comparing it with those of TDM / FDM systems.

Statistical analysis of computer generated holography for data compression applications

ABSTRACT. Holography is an acquisition and reproduction technique of visual contents which allows, theoretically, the reconstruction of the acquired scene without any difference with its real-world counterpart. Digital holograms are characterized by a huge amount of data, having different features compared to conventional image and video content. This paper presents a statistical analysis of computer-generated holograms data with the aim of providing a better understanding of the features and characteristics of holographic data, to be used as a basis for research and design of novel coding solutions specifically tailored for holographic contents.

Performance Analysis of Chaotic Symbol Dynamic Decision-based 8-APSK with Enhanced Security for Broadcasting Applications

ABSTRACT. This paper is on the analysis of BER performance of 8-APSK using a decision rule called chaotic symbol dynamics that can adjust the decision region of signal constellation towards optimum BER and security enhancement of the transmitted multimedia. In this regard, an iterative chaotic map is utilized for 8-APSK as a partitioning tool of decision regions to dynamically adjust the minimum Euclidean distance between two successive symbols with a control parameter p. Analytically calculated BER using the proposed decision rule and simulation results show that this decision rule based chaotic 8-APSK can yield almost identical performance to 8-APSK (coded and uncoded cases) with minimal degradation while effectively prevent unauthorized demodulation without prior knowledge of parameter p, hence thus maximizes the information security.

A DRL-based resource allocation framework for multimedia multicast in 5G cellular networks

ABSTRACT. Abstract—Cloud Radio Access Network (C-RAN) is a key enabling network architecture for the next generation (5G) wireless communications and has advantages of providing multimedia service on the basis of properly resource allocation. However, the resource allocation strategy still needs to be further improved in order to allocate resource where unicast or multicast multimedia services are requested simultaneously. Aiming at minimizing power consumption and meeting quality of service (QoS) requirement of users, this paper proposes a deep reinforcement learning based unicast-multicast resource allocation framework (DRL-CRAF) for it with high energy efficiency. Specifically, we define the state space, action space and reward function for the DRL agent, and formally formulate the resource allocation problem in a multimedia service scenario as a convex optimization problem. We evaluate the performance of the proposed framework by comparing it with three widely-used baselines via simulation. Simulation results show that the proposed method can not only meet the requirements of both unicast and multicast users but also effectively reduce 10% transmission energy consumption compared to full connectivity method.

Dual Convolutional Neural Network for Image Steganalysis

ABSTRACT. Image steganography is a covert communication technique that embeds secret message into cover images and makes the embedded message not perceived by the others. On the contrary, when a suspicious image is given, image steganalysis is a technique that identifies whether steganography was applied to the image. Steganalytic algorithms are classified into handcrafted feature based and convolutional neural network (CNN) based ones. Handcrafted feature based steganalytic algorithms extract statistical features from images and learn them using machine learning algorithms to identify message-embedding images (called stego images). Their performance strongly depends on how to design and extract features. On the other hand, CNN-based algorithms automatically extract/learn the features optimal for the identification. Basically, CNN can be used for image steganalysis in the same way as for the conventional classification problems such as object detection/recognition. However, to use CNN in steganalysis, additional preprocessing filter is required to amplify subtle changes between before and after embedding secret messages. Therefore, most CNN-based steganalytic algorithms have focused on designing more effective preprocessing filters or CNN structure. Unlike the common approach, in this paper, we propose a new approach that uses dual CNN of which each has different inputs. To construct the dual CNN structure, two pairs of the preprocessing filters and the convolutional layers were brought from the conventional CNN-based steganalytic algorithms and their outputs were fed together into a following affine layer. Some additional data is embedded into the input image by using one of steganographic algorithms and a difference image is computed between the input image and its stego image. Then, two different images, the input and difference images, are fed into each CNN, respectively. This indicates that the proposed approach extracts/learns additional features from the difference image using the additional CNN. This is based on the fact that the pixel variations occurred by embedding data into a cover image are different from those occurred by embedding data into a stego image. For experiment, 10,000 gray scale images of 512 x 512 in BOSSBase 1.01 were quartered and the resulting 40,000 images were used as cover images. Stego images were generated using the S-UNIWARD (one of content-adaptive steganographic algorithms) with a fixed key and payload = 0.4. For the proposed method, additional data was embedded into the cover and stego images using the same steganographic algorithm with a fixed key and payload = 1.0. Experimental results are as follows. The accuracy was better when the same CNN was used for dual-CNN than when two different CNNs were used. Also, the accuracy strongly depended on the optimizer used for learning each CNN. The accuracy was better when the same optimizer was used for both CNNs than different optimizers were used. When the different optimizers were used, the accuracy was worse than the conventional steganalytic algorithm. Therefore, the same CNN and optimizer were used for configuring the proposed dual CNN. Consequently, the proposed dual CNN with additional input could identify whether steganography was applied to the input image with an accuracy of 80%, and could improve the accuracy by approximately 5% when compared with the conventional CNN-based steganalytic algorithm.

Comparison of FDM with LDM on hierarchical transmission toward Advanced ISDB-T development
PRESENTER: Akihiko Sato

ABSTRACT. Japan Broadcasting Corporation (NHK) is conducting research on the Advanced Integrated Services Digital Broadcasting-Terrestrial (Advanced ISDB-T) system to enable Ultra High Definition (UHD) content, such as 4K and 8K on terrestrial television broadcasting system[1][2]. We are considering upgrading a new signal structure, carrier modulation, forward error correction (FEC) of the current ISDB-T system. In addition, same as current ISDB-T, we have been considering the adoption of frequency-division multiplexing (FDM) as its hierarchical transmission system to multiplex both fixed and mobile reception services. However, we also research on layered division multiplexing (LDM) as a hierarchical transmission system that can improve the spectrum efficiency. We therefore developed prototypes of a modulator and demodulator that are compatible with both FDM and LDM to which the new signal structure, carrier modulation, and FEC have been applied. We show the result of evaluation and comparison of transmission capacity between FDM and LDM in this paper.

Methodology for Analyzing ATSC 3.0 Single Frequency Network Overlapped Area Based on Spatial Interpolation
PRESENTER: Sungho Jeon

ABSTRACT. In extending the single frequency network (SFN), it is inevitable that the newly added broadcasting stations are established while the existing broadcasting stations are being operated. To identify the SFN signal overlapped area before and after the construction and to check whether the newly added transmitter signals are stably inter-operable to those of the existing transmitters without self-interference, the measurement vehicle should be fixed at a certain measurement point and the transmitter must be turned on and off up-to-date. Hence, as the number of fixed measurement points increases, the transmitters will be exposed to severe stress. Additionally, because the fixed measurement takes a long time to collect enough volume of data, it is not practical to use the old technology as it is. In this study, we propose a novel method to collect the field test data by mobile measurement under the fixed transmitter condition, i.e., turned on or off. Then, the difference between the two results is calculated using data processed by spatial interpolation to determine the SFN signal overlapped area quickly. After applying this to field test data, it is confirmed that coverage of each transmission site obtained by the proposed method is almost identical to that of the coverage prediction and planning software.