ABSTRACT. Today's most widely used broadband communication systems are based on Orthogonal Frequency Division Multiplexing (OFDM) modulation format due to its numerous advantages (e.g. resistance to multipath propagation, high spectral efficiency). However, OFDM is very sensitive to In-Phase and Quadrature (IQ) errors in the transmitter and receiver front-ends. This paper deals with the study IQ-error influence on the performance of te second generation digital video broadcasting standard (DVB-T2). Ricean and Rayleigh fading channel models with 12 echo paths are used to emulate a rooftop reception scenario. Furthermore, three low complexity methods to suppress different IQ-errors are tested and compared. Bit error rate depending on carrier-to-noise ratio is used to describe the performance of DVB-T2 system influenced by IQ-errors. Results reveal that a DVB-T2 receiver can successfully eliminate lower level of IQ-errors without compensation techniques.

ABSTRACT. Broadcast systems based on FDM (Frequency Division Multiplex) have the advantage of near continuous demodulation of the broadcast signal, allowing accurate and continuous tracking of channel conditions which is particularly useful for mobile reception. This has been employed in the ISDB-T standard used in Japan, Brazil and other countries. However, as designed in ISDB-T the broadcast signal lacks the ability to send system parameters such as FFT size, GI size and so on before the receiver begins demodulation. The receiver must blindly estimate such system parameters before it can read the other detailed parameter information using the TMCC pilot carriers. This takes time, usually one frame or longer. This paper proposes a next generation FDM system which enables the original advantages of FDM to be retained, while allowing additional advantages by employing an additional small signal (Preamble 1) which imparts essential information such as FFT size, GI size and pilot pattern to the receiver to enable immediate demodulation of the broadcast signal based on known parameters rather than blind estimation. Following demodulation of the first preamble, demodulation of the second preamble (Preamble 2) allows immediate knowledge of the all subsequent parameters contributing to faster demodulation of the overall signal.

ABSTRACT. Transmit diversity code filters set (TDCFSs) adopted in ATSC3.0 is a method using all-pass linear filters to minimize the possibility of cross-interference among the transmitted signals. This paper presents a new cyclic algorithm for the filter set design in TDCFS. In the case of designing the filter set with almost identical correlation property, the proposed algorithm requires less iteration times than the algorithm from the literature because of its simplicity in calculation. Thus, it can be used to design the filter set consisting of a huge number of filters and long sequences filters with relatively lower complexity compared with the algorithms in literature. A numerical example and system simulation results are also provided in this paper to demonstrate the better performance than that of the state of the art.

ABSTRACT. The new broadband frontier for satellite communications will be the exploitation of the bandwidth portions in the millimeter wave domain, namely: Extremely High Frequencies (EHFs). Some evident advantages can be taken by the usage of EHFs for satellite broadcasting and, more in general, multimedia satellite services. However, some open issues are still to be solved to translate such theoretical advantages into actual service offer. The present work aims at providing a survey about opportunities and challenges of EHF exploitation for satellite services to residential customary, including high-definition TV broadcasting, interactive multimedia applications, emergency communications etc. The future perspective of the integration with terrestrial 5G that will exploit EHF for high-capacity transmission over small cells will be considered as well.

ABSTRACT. A novel motion-compensated frame rate up conversion algorithm is proposed in this paper, which employs the iterative non-local means (NLM) interpolation. Firstly, we obtain the forward and backward motion vector fields (MVFs) by optical flow estimation. Then in order to reduce the cracks and salt-and-pepper noise, the initial interpolated frame is reconstructed by the fusion of these two MVFs in block-based interpolation. After that an iterative non-local means interpolation is proposed, where the MVFs and intermediate frame are refined alternatively in each iteration. This iterative procedure can alleviate the artifacts and preserve the texture effectively. Experimental results demonstrate that the proposed method outperforms the state-of-the-art algorithms in both objective and subjective quality.

Frame rate up conversion (FRUC) is widely used to improve the visual quality by converting low frame rate video into high one. Recently, the rapid growth in multimedia and display devices has lead to the demand for FRUC in many applications, such as high-end television and mobile devices.

Most FRUC algorithms are based on motion-compensated frame rate up conversion (MC-FRUC). A typical MC-FRUC consists of two procedures: motion estimation (ME) and motion compensation interpolation (MCI). Block matching algorithms are commonly used for motion estimation \cite{dikbas2013novel,liu2012multiple,kim2014new,guo2014frame} due to its straightforward and hardware-friendly implementation. However, the block matching algorithms do not provide adequate performance in correctly estimating all the motion in a frame. In \cite{kim2014new,guo2014frame}, motion vector refine (MVR) methods have been introduced. However, the estimated MVs are still not reliable enough and may produce blocking artifacts in interpolated frame. An alternative solution is to utilize the optical flow estimation, which is a pixel-level motion estimation method. In \cite{lee2014frame}, Lee \textit{et al.} generated four intermediate frames by optical flow and reconstructed the final interpolated frame in variational framework. Since the intermediate frame is interpolated in pixel-level, it may suffer from salt-and-pepper noise \cite{tang1998comparison}. %In \cite{kavianiframe}, optical flow estimation and patch-based reconstruction were employed to obtain the interpolated frame. Although the noise and cracks are reduced, the ghost artifacts and over-smoothing may exist in motion boundary due to the averaging of all stacked values in same position.

In addition to motion estimation, a lot of researches have been carried out to improve the quality of interpolated frames. In \cite{choi2007motion}, Choi \textit{et al.} proposed the adaptive overlapped block motion compensation to reduce the blocking artifacts. In \cite{wang2010frame}, Wang \textit{et al.} employed the trilateral filter to preserve the edges and restore the missing pixels. Since the performances of these MCI methods are highly dependent on the reliability of MVs, the artifacts and uneven texture may still exist in complex motion. In \cite{dehghannasiri2014frame}, Dehghannasiri \textit{et al.} employed a nonparametric estimator in which a number of pixels in the adjacent frames of the intermediate frame are contributing to the interpolation. However, they do not perform motion estimation and the intermediate frame is reconstructed by the linear combination of all blocks in the search window, which may lead to over-smoothing and loss of texture. Besides, the occlusion problem is not discussed in \cite{dehghannasiri2014frame}.

In this paper, a novel FRUC algorithm which employs the iterative non-local means interpolation is proposed. In order to obtain reliable interpolated frame, the forward and backward MVFs obtained by optical flow are projected into the intermediate frame, and the initial interpolated frame is reconstructed by the fusion of these two MVFs by block-based interpolation. More importantly, we improve the quality of the initial interpolated frame further by an iterative NLM interpolation scheme. In each iterative process, both the intermediate frame and MVFs are refined. Specifically, we obtain the similar blocks that indicted by spatial neighbour MVs and combine them together in a NLM approach to refine the initial interpolated frame. Then the refined frame is in turn used to update the MVFs. These iterative processes can effectively reduce the artifacts step by step. Experimental results demonstrate that the proposed method provides better interpolation performance than the existing methods.

The rest of paper is organized as follows. In Section \ref{sec:ME}, the reconstruction of initial interpolated frame is discussed. Then the proposed iterative NLM interpolation is detailed in Section \ref{sec: NLM interpolation}. Section \ref{sec:experimental} presents the experimental results. Finally, we conclude the paper in Section \ref{sec:conclusion}.

ABSTRACT. Steganography is a technique for covert communication where a secret message is not perceived by the third person between a receiver and a transmitter. It has been developed for thousands of years for the transmission of military, diplomatic or business information. The advent of digital media and communication has led to the development of digital steganography techniques in modern times. Image steganography is one of the digital steganography techniques. Image steganography embeds secret information into the spatial domain or the frequency domain of images. For example, the pixel value differencing algorithm is used to embed data into the spatial domain, and discrete wavelet transform (DWT) or discrete cosine transform (DCT) are used to embed data into the frequency domain. In case of DCT, secret data is simply embedded into the DCT coefficients. DWT generates LL, LH, HL, and HH sub-bands. Each sub-band represents a band passing LPF and LPF, HPF and LPF, LPF and HPF, and HPF and HPF in horizontal and vertical directions. In DWT-based image steganography, secret data is mainly embedded for LH, HL, and HH sub-bands. This paper proposes image steganography which embeds secret data using block matching and DWT. A cover image is transformed by integer wavelet transform (IWT) algorithm to embed secret data in the frequency domain. IWT divides the cover image into LL, LH, HL, and HH sub-bands. LL sub-band contains low frequencies, and LH, HL, and HH sub-band contains high frequencies. The LL sub-band is used as a database for block matching. LH, HL, and HH sub-bands are less perceptible by data modification than the LL sub-band so they are used for embedding secret data. To perform block matching between the cover and secret image (assuming that the secret data is also an image), the LL band of the secret image is divided into n by n non-overlapping blocks and that of the cover image is divided into n by n overlapping blocks. For a block of the secret image, its closest block is searched among the blocks of the cover image. It might not be possible to find the similar block due to the different illumination condition, so the values of the blocks are normalized based on their minimum. The mean square error (MSE) is used for computing the similarity between blocks, and a block with the smallest MSE is selected. Once the closest block is found, its location (x-coordinate and y-coordinate in the cover image) is embedded into two least significant bits of the LH, HL, and HH bands of the cover image. In the decoding phase, the secret image cannot be recovered by 100% (because it is composed by the similar, not the same, blocks found in the cover image). Thus, the peak signal-to-noise ratio (PSNR), which represents the similarity between the original secret image and recovered image, is used to evaluate the recovery performance of the proposed method. Also, the PSNR between the cover image and the stego-image is used to evaluate the visual perceptibility of the data hiding by the proposed method. In the experiments for several cover images and secret images with the same resolution of 512 by 512 pixels, the block size of 2 by 2 showed good performance with the PSNRs higher than 30 dB and the embedding capacity of around 100 kB. By using the block size of 4 by 4, the PSNR became lower but the embedding capacity increased to around 262 kB. Therefore, it was necessary to determine an appropriate block size for the secret image, to balance the PSNR and embedding capacity.

ABSTRACT. In this paper, we propose a novel electronic scissoring algorithm for image detachment. Our proposed method relies on the pre-segmentation and hence it is computationally efficient. Our proposed feature is computationally simple and robust over different object and text sizes. Our proposed electronic scissoring algorithm contains two major steps, namely informative content extraction and image-content detection, to eliminate the non-image content, which belongs to the pure text only. For the informative content extraction, a variance filter is employed to serve for this purpose. For the image-content detection, the entropy measure is used to quantify the randomness of an extracted informative content. Generally speaking, image contents appear to be more complicated than the texts or other non-image patterns in terms of pixel-intensity distributions. Therefore, the entropy can best distinguish the image and non-image contents. Experiments are carried out to evaluate the performance of our proposed algorithm, which demonstrate the corresponding effectiveness.

ABSTRACT. In this paper, we propose an enhanced projector using the emergent display technology of Temporal Psycho-Visual Modulation (TPVM) which exploits the difference in image formation mechanisms of human eyes and imaging sensors. The projector have two significant advantages compared to traditional projector: 1) parts of the information (e.g., speech notes) projected can be hidden from the audience but the speaker can see the hidden parts; 2) the speaker's eyes can be protected from exposing to the light emitter directly. In the proposed system, every projected image is split into two subimages, where one contains both the original and hidden information, and another is opposite to the first subimage to achieve the purpose of hiding some information. To avoid the harmful stimulation from projection light to eyes, the enhanced projector detect the position of the speaker's face and adaptively modify the subimages to protect the speaker's eyes from directly exposing to the light source. The proposed prototype system uses the traditional projectors which support 3D display mode, and the proposed eye-protecting and information-hiding scheme is implemented by software. So it is transportable and practicable.

ABSTRACT. In the ATSC 3.0 PHY layer standard, a Transmitter Identification (TxID) signal is defined in order to provide the identification of an ATSC 3.0 transmitter. For efficient use of spectrum and quality of service improvement, single frequency network (SFN) where all transmitters share a single RF channel is often implemented. The recently developed ATSC 3.0 physical layer standard has been designed to support SFN. For efficient designing, fine-tuning and operating an SFN, it is crucial to identify each transmitter, as well as to adjust transmitting power and emission time of each transmitter. This paper presents TxID for ATSC 3.0, and analyzes its detection performance under very low signal to noise ratio (SNR) environments.

ABSTRACT. This paper proposes a new transmitter identification mechanism for DVB-T2 based on direct sequence spread spectrum (DSSS) modulation. The performance of the proposed method will be compared with the solutions included in the current version of the standard. Finally, the potential inclusion on the transmitter identification of new capabilities, such as the Emergency Alert System (EAS) will also be considered.

ABSTRACT. This paper describes an integrated system of alert and alarm emergency, which uses robust radio diffusion technologies recommended by ITU (Digital Terrestrial Television and FM Radio). For the insertion of the emergency information in the radio frequency, generators were developed for FM radio and television using RDS (Radio Data System) and information tables respectively. A receiver module developed by INICTEL-UNI (Instituto Nacional de Investigación y Capacitación de Telecomunicaciones ) decodes the emergency signal and activates sound sirens to alert target audience. This emergency signal is based on the CAP protocol (Common Alerting Protocol) and PLANAGERD (Plan Nacional de Gestión del Riego de Desastres), which allows defining multiple emergency events for dissemination. Finally the management of the system is carried out through a web interface, which is responsible for the emergency activation of multiple events and control and monitoring the receiver module parameters.

ABSTRACT. Since FTN signaling uses greater symbol rate than Nyquist rate, the FTN signaling offers higher spectral efficiency than that of Nyquist signaling. However, ISI is derived from the non-orthogonality of FTN signaling. The ISI leads to an increase of the envelope fluctuations. Consequently, high PAPR occurs in the FTN signaling. In the satellite communication link, the high PAPR causes performance limitation because of non-linear amplification effects. Therefore, to adopt FTN signaling, PAPR reduction scheme is required in satellite digital broadcasting system. Furthermore, for various reasons such as security of pay-contents, data masking can be efficiently used for digital broadcasting system. Therefore, this paper proposes joint masking and PAPR reduction technique based on scrambling signal frame for the broadcasting system based on DVB-S2 standard with FTN signaling.

ABSTRACT. NHK has been conducting research on high-order-modulation orthogonal frequency-division multiplexing (OFDM) schemes such as 4096QAM, dual-polarized multiple-input multiple-output (MIMO), and low-density parity-check (LDPC) code schemes, for large-capacity transmissions of next-generation digital terrestrial broadcasting systems. In 2014, we conducted a field experiment on MIMO-OFDM transmission in Hitoyoshi-city, Kumamoto prefecture, Japan. In 2016, the technologies were tested in a field experiment in Rio de Janeiro (RJ), Brazil. In addition, we conducted a transmission simulation of MIMO-OFDM and a calculation of the MIMO channel capacity using the deteriorated MIMO channel measurements from RJ. In our previous study, we reported that using weighted log likelihood ratios (LLRs) and an erasure process in each subcarrier reduces the bit error rate (BER) in multipath environments. The erasure process sets the LLR to 0 when the MIMO channel of a subcarrier deteriorates. In this paper, we examined the BER performance in cases with or without the erasure process using the deteriorated MIMO channel measurements from RJ. As a result, even when the MIMO demodulator uses a zero forcing (ZF) algorithm with a higher order modulation such as 4096QAM, we got quasi-error free (QEF) results when using the erasure process.

ABSTRACT. Advanced Television Systems Committee (ATSC)3.0, which is the new international standard for next generation broadcasting, allows to deliver multiple services with different reception conditions in a single RF channel. In the physical layer standard of ATSC 3.0, two kind of multiplexing schemes are adopted: orthogonal schemes including traditional time division multiplexing (TDM) and frequency division multiplexing (FDM), and a newly introduced non-orthogonal scheme, named layered division multiplexing (LDM). In this paper, we consider two services configured by LDM and TDM based on the ATSC 3.0 standard. Field test results which evaluate the mobile performance of LDM and TDM are provided.

ABSTRACT. In ATSC 3.0 physical layer standard, a layered division multiplexing (LDM) scheme has newly accepted as one of the multiplexing schemes in order to improve the reception performance when multiple broadcasting contents are simultaneously delivered. This paper presents comparison results of LDM and time division multiplexing (TDM) schemes under fixed reception environments, and analyzes the performance gain of LDM over TDM.

ABSTRACT. This paper presents the preliminary results of a field measurement campaign for the digital terrestrial television DVB-T2 signal in Seoul. This includes an analysis of its horizontal and vertical variations as well as the major link budget factors in planning digital terrestrial television networks for indoor reception, such as height loss and standard deviation of the location correction factor. By modeling the values measured from the obtained parameters as a log-normal distribution, the efficacy of indoor service coverage can be predicted from the target data rate.

ABSTRACT. Alpha matting is known as a process of accurately separating a foreground object from an image and it has been extensively studied in both academic and industrial areas as an important part of multimedia technology. However, most existing methods utilize single image to get its alpha matte. In this paper, we propose a novel chroma keying method based on stereo images, which can be applied to post-process the alpha matte generated by any matting approach. Given a pair of stereo images, a new matting Laplacian matrix is constructed based on the affinities between matching pixels and their neighbors from two frames. Based on the new matting Laplacian matrix, a new cost function is also formulated to estimate alpha values of the reference image through the propagation between stereo images. Experimental results illustrate that this method can be applied to different categories of current alpha matting approaches and can obtain alpha matte of higher quality.

ABSTRACT. An iterative-weighted-average algorithm for background modeling in surveillance video scenarios is proposed in this paper. Through allocating the pixels at the same position in several continue frames in a time window with different weight values during the iterative process, the influence of foreground pixels on background modeling can be removed immediately without requiring massive hardware memory. Furthermore, the algorithm we propose can also perform well in time-varying backgrounds such as the parking lot. Simulation result demonstrates that the proposed algorithm is capable of saving a large amount of coding bits, while maintaining the same computing complexity and SNR.

ABSTRACT. In this paper, we present a real-time logical screen composition scheme which focuses on automatically detecting and tracking quadrangles for screen region. It has novel concept that the detected region is created as a logical screen so that any non-planar region can be designated as various digital signage spaces. In this paper, we also present service scenarios and show service feasibility through implementation. The proposed scheme is expected to help realize various digital signage broadcasting services.

ABSTRACT. In this paper we propose a novel but simple spatial invariant convolution neural network for crowd counting. Our model combines the spatial transformer unit with the CNN net aiming at addressing the counting task under complicated scenes with server spatial variant. There are three mainly privileges of our method. Firstly, our model can improve the CNN net with little cost while the net is easy to train. Essentially, we try to enhance the CNN net itself rather than merely changing the architecture. Secondly, our model does not need the pooling layer which can deduce the systematic error caused by the resize of the ground-truth density map. Thirdly, our method can been easily transplanted to different scenes due to the powerful spatial invariant learning ability of our model. Furthermore, we have collected a dataset with challenging scale and pose variant of crowd with about 20 videos . The preliminary experiment result shows the practicability and effectiveness of our method.

ABSTRACT. Combining Spatial Modulation (SM) with multiple input multiple-output (MIMO) systems constitutes a promising technique for exploiting full spatial multiplexing. To further maximize its capacity, we propose to optimize the activation probability with an iternative optimization method. Moreover, we propose to incorporate the newly-proposed distribution matcher encoder for probability shaping. Numerical results are provided to substantiate the proposed system’s improvement in terms of capacity, and to validate the efficiency of the proposed encoding scheme

ABSTRACT. ATSC 3.0, the next-generation U.S. Digital Terrestrial Terevision (DTT) standard, allows a higher spectral efficiency and/or a transmission robustness with Multiple-Input Multiple-Output (MIMO) technology compared to existing DTT networks. Regarding MIMO channel estimation, two pilot encoding algorithms known as Walsh-Hadamard encoding and Null pilot encoding are possible in ATSC 3.0. This paper focuses on the performance evaluation of the two MIMO pilot encoding algorithms in ATSC 3.0 using physical layer simulations. There are no published articles in the literature covering this topic yet. This paper provides a great benefit to broadcasters to select the MIMO pilot configuration that better suits their service requirments. Several channel estimation algorithms have been evaluated in both mobile and fixed reception conditions. The preliminary simulation results show that Null pilot encoding provides slightly better performance than Walsh-Hadamard encoding for fixed reception but worse performance for mobile reception, especially at high signal-to-noise ratios.

ABSTRACT. A real-time high-speed visible light communication (VLC) system based on RGB-LEDs is proposed in this paper. With the use of the wavelength-division multiplexing (WDM), the system enables three individual color channels. The physical layer of the system is based on time-domain synchronous OFDM (TDS-OFDM) technology. Digital pre-distortion method also is applied to improve the SNR performance. Experiment demonstration shows that the proposed system can achieve a peak transmission data rate of 544.32 Mbps in real time with 256-order constellation, transmitting three individual 4K ultra-high-definition television (UHDTV) signals at the same time. The proposed system is proven to be sufficient for wireless indoor multimedia communication requirements.

ABSTRACT. The compatibility of Geneva 2006 frequency plan
(GE06) assignments and allocations for digital terrestrial television
(DTT) is determined using ITU-R P.1546 propagation
prediction method. The aim of this paper is to study the suitability
of using ITU-R P.1546 propagation predictions for allocating
Long Term Evolution (LTE) Supplemental Downlink (SDL) to
broadcast Audiovisual Media Services (AVMS) in the UHF TV
band. This paper presents a field measurement campaign to
study the accuracy of P.1546 propagation predictions for LTE
base station signals in the close vicinity of the LTE base station,
where DTT reception interference and overloading events are
most likely to occur. The ITU-R P.1546 prediction signal strength
in general was higher than the measured field strength, which
means that using ITU-R P.1546 propagation predictions could
thus overestimate the probability of interference events to DTT
reception. The paper also studies use of more sophisticated
site-specific propagation prediction methods, but the considered
short-range propagation paths were found to be very difficult
to predict regardless of the used method. The location-bylocation
analysis indicates that the propagation predictions are
less accurate in the locations where the line-of-sight (LOS) or
first Fresnel zone are obstructed.

ABSTRACT. See attached.

ABSTRACT. The Advanced Television Systems Committee (ATSC) 3.0 is the latest digital terrestrial television (DTT) standard. It offers, among others, unprecedented features and flexibility for ultra-robust transmission with Carrier-to-Noise ratios (CNRs) below 0 dB. Negative CNRs may cope with degradation caused by e.g. noise, Doppler shift, Inter-Symbol Interference (ISI) by multipath or even self-interference in Single Frequency Networks (SFNs). The latter may enable the suppression of a guard interval (GI), providing high gains in the context of SFNs where long GIs that represent important overheads are required. This paper studies the performance of ultra-robust transmission modes of ATSC 3.0 without GI in SFNs. A set of these modes are compared in several SFN scenarios, characterized by different power imbalances and path delays. The impact of the GI duration, scattered pilot (SP) patterns and channel estimation algorithms are also analysed. The initial results confirm that at low rates (QPSK modulation with code rate 2/15 and 3/15) the effect in performance of eliminating the GI is less that the capacity overheads in SFNs, what may increase transmission efficiency in such Zero-Guard OFDM systems.

ABSTRACT. It is worthwhile to study the impact of terrestrial network coverage on television media markets because the terrestrial broadcasting is considered as the most influential media platform mainly due to its wide network coverage. However, it is not clear how the coverage contributes to the media's prominence. To understand its impact, we focus especially on its economic impact on the market which consists of terrestrial broadcasters and pay TV operators. The pay TV operators retransmit the terrestrial TV channels under a retransmission consent, so the broadcasters can be seen as vertically integrated operators which are both upstream TV program rights holders and downstream rival distributors to the pay TV operators. By dividing the market into 'inside the coverage' and 'outside the coverage', we investigate the effect of the coverage on subscription price of the pay TV, and profits of both firms. We also study the change of the retransmission fee since the coverage changes their profits thereby affecting the bargaining power.

ABSTRACT. A light field image is a sampling of the intensity and direction information of the light rays crossing the main lens of a digital light field camera.
The light field information captured by a camera must be processed in order to render images of the scene to the final user at a given focal plane or viewpoint.
In the area of light field imaging, several compression techniques are currently under research in order to foster the development of novel applications.
An analysis of raw light field image compression of sliced lenslet array is presented in this paper. The distortion introduced by compression is measured in terms of peak signal-to-noise ratio and the structural similarity index.
The obtained results are finally discussed.

ABSTRACT. In this paper, we present a new group of pictures (GOP) level rate control bit allocation method for High Efficiency Video Coding (HEVC), which is the state-of-the-art video coding standard from ITU/ISO. Because of the difference of GOP coding structures between HEVC and H.264/AVC, the GOP level bit allocation method of H.264/AVC rate control is no longer adapted to HEVC. However, the existing rate control algorithms do not address it correctly. Thus, we propose a new GOP level bit allocation method which matches HEVC GOP coding structure exactly. Experimental results show that the proposed GOP level bit allocation method can achieve more accurate rate control, lower bit fluctuation, and slightly better RD performance than the original rate control algorithm in HM-16.9 (HEVC reference software) in both random access and low delay configurations without increasing encoding complexity.

ABSTRACT. VP9 is an open source royalty free codec developed by Google for web video applications, aiming at competing with current mainstream standard HEVC in compression efficiency and quality. Both VP9 and HEVC have seen increasing market acknowledgement 3 years after their completion since growing number of leading companies join to support them, especially VP9. So the conversion between VP9 and HEVC is in need. However, transcoding between different standards is time consuming as the codecs become more and more complex, which brings the demands for algorithms to make the transcoding process faster without severely damaging the quality. In this paper, we explore techniques and algorithms based on machine learning to find the patterns and connections between VP9 decoded information and HEVC encoding decisions, and thus accelerating the transcoding process from VP9 to HEVC. According to the experiment, a maximum time saving of 53% and an average of 44% can be achieved with inevitable yet acceptable quality loss.

ABSTRACT. High Efficiency Video Coding (HEVC) is the latest video coding standard established, which introduces a number of new coding tools such as hierarchical structures of coding unit (CU), prediction unit (PU) and transform unit (TU) to improve the compression capacity of coding. Compared with H.264/AVC, the compression performance increased by nearly 50% at the cost of a significant increase in algorithm complexity. One of the most time-consuming parts is the inter coding process which takes over 60% of the encoding time. In this paper, in order to reduce the computational pressure on the encoder side, we design an algorithm based on spatio-temporal characteristics of the encoding pictures. First, we adopt a CU depth pruning strategy based on the texture feature and motion information of the blocks. Then, based on the motion vectors (MVs) of adjacent coded blocks, we introduce a fast reference picture selection strategy. Experimental results show that the proposed algorithm achieves 45% time saving on average with negligible loss of coding efficiency compared with HEVC reference software HM16.9.

ABSTRACT. Cognitive Radio (CR) technology was introduced to answer the spectrum scarcity problem by enabling unlicensed users (secondary users) to access opportunistically the licensed spectrum bands, when the licensed users do not transmit. This paper work introduces a broadcast protocol for Multi Hop Ad Hoc Networks that organizes secondary users (SUs) in clusters in a distributed manner based on the locally available channels with a dynamically selected Common Control Channel (CCC) for the cluster’s operation and reliable broadcast delivery. When a broadcast message is sent on the CCC, it reaches all the nodes of the same cluster and is transmitted immediately from the cluster’s edge nodes to the neighboring cluster’s CCC i.e. the neighboring group of SUs; all broadcasting transmissions have guarantees for Collision Avoidance, low latency, no additional delay. To the best of our knowledge, there is no other work with such guarantees and bandwidth assurance provision. The clusters are self-organized such that they can be cloud-assisted on demand. Simulation results present a comparative study with multi-hop broadcasting protocol for CR Ad Hoc Network.

ABSTRACT. I. INTRODUCTION The exponential growing of mobile devices at increasing reduced costs, is leading to a significant increment in the number of users that need broadband multimedia system and relative radio resources. Nowadays, there is the need to be connected to the network everywhere and anytime according to the paradigm ABC (Always Best Connected) [1]. Moreover, technological improvements, a wide offer of multimedia contents and an ever-growing distribution of mobile applications represent an unrestrainable trend. Video traffic will account for more than 69% of mobile data traffic by 2018 and, in 2015 the total mobile network traffic has been account for more than 65% of the total network traffic, mainly concerning the video traffic (i.e., streaming, real-time applications, TV on-demand, etc) [2]. In order to improve quality of service (QoS) network operators overcome with this problem deploying several radio access technologies (RATs) to the radio network type [3] Different solutions offer to the users the maximum QoS through architectures and algorithms that provide a dynamic vertical and horizontal handover, power received maximization, energy savings, and ensuring at the same time costs minimization bonded to their own mobile network operator agreement. The presence of such a substantial heterogeneity of available technologies makes critical the selection of the best candidate network ensuring the higher QoS, and avoiding network saturation that would significantly decrease an acceptable user experience. One of the possible solutions is performing the traffic balancing across heterogeneous networks, making sure that the available networks for a given user at a given time are not overloaded. Currently, most of the load balancing algorithms carry out the network selection by combining the power of the received signal, the throughput, the packet delay, the monetary cost for the user, and the energy consumption. This work proposes a novel algorithm to select the candidate network based on the Multiplicative Exponent Weight method (MEW) [4] for an improved real-time balance of available radio resources.

II. RELATED WORKS Different approaches for the selection of candidate networks in order to obtain a network load balancing have been proposed. In [5], the authors proposed a solution that makes use of a reputation-based mechanism to select the most appropriate set of networks (for the mobile user) and a load balancing mechanism to distribute the traffic load among the networks by making use of the Multipath TCP (MPTCP) protocol. In [6], the authors propose a user location-aware reputation-based network selection solution that aims at improving the video delivery in heterogeneous wireless network environment by selecting the best network score. In [7], the authors focused on the realization of a heterogeneous network environment with a combination of macro-cells and small cells to spread the traffic load, increase the bitrates and maintain the high QoS. In [8] the topic is the selection of candidate networks based on the received signal strength. Using the repeated Prisoner’s Dilemma game, it models the interaction with the user-network as a cooperative game and shows that defining the incentives and disincentives for cooperation against defection on service guarantees, repeated interaction supports cooperation. These types of algorithms do not take into account the required type of traffic at that time, the type of device being used and the user priority defined as the capability to satisfy user requests compared to others. To overcome these limitations, in [9] authors propose an adaptive real-time multi-user access network selection algorithm for load-balancing over heterogeneous wireless networks (ARMANS) through the definition and the use of four different QoS classes, as defined by the Wi-Fi standard [10]. ARMANS improves the balance of the networks than studies that preceded it. This fact motivates our work to provide a better solution compared to ARMANS algorithm. The proposed MEW solution considers a scenario in which a wide range of heterogeneous transmitters is employed in a real scenario as proposed in Figure 1. Compared to ARMANS algorithm, MEW also introduces the Signal-to-Interference-plus-Noise Ratio (SINR) and monetary cost to select the candidate network.

III. MEW ALGORITHM The MEW method is used to determine the score of each network to which the user can access, through the multiplicative mathematical operations. Alternative criteria are normalized with the relative weights, according to the different relevance that is attributed to them. The utility function applied in this study is given by the equation (1): U^i=〖(u_l^i)〗^(w_l )∙(u_p^i )^(w_p )∙(u_s^i )^(w_s )∙(u_v^i )^(w_v )∙(u_c^i )^(w_c ) (1) where U^i represents the overall value of the utility function that considers five different criteria for the i-th network.u_l^i represents the utility function for the load, u_p^i is the signal strength, u_s^i is the SINR, u_c^i is the monetary cost of the network, and u_v^i represents the user mobility. The weights〖 w〗_l, w_p, w_s, w_c, and w_v, represent the normalized (i.e., [0-1]) incidence of load, signal strength, SINR, monetary cost and user mobility, respectively. The network with the higher score U^i represents the best candidate network. Simulation test-based are validated studying a urban real scenario of the pedestrian area in front of the coast of Poetto (Cagliari, Italy) as shown in Figure 1.

The preliminary simulations results show a better performance of MEW when increasing the number of users. For instance, Figure 2 shows the MEW and ARMANS curves of PSNR while increasing the number of users. In particular, MEW algorithms shows constant PSNR for a higher number of users compared to ARMANS curve. After saturation point, the MEW curve shows a slower decrease performance than ARMANS. In the full paper, a study on the influence of different parameters of QoS and the performances in terms of PSNR, average PSNR, throughput, aggregate throughput, delay, and packet loss will be detailed.

IV. CONCLUSIONS In this work, we propose a novel algorithm that performs a real time load balancing over heterogeneous wireless networks using a dynamic MEW function based on the user's characteristics. The results obtained in the simulation phase, show an improvement in performance in terms of QoS evaluated in terms of PSNR, average PSNR, throughput, aggregate throughput, delay, and packet loss. It has achieved an average improvement in all metric of 15% respect to the balancing performed with the ARMANS.

ABSTRACT. I. INTRODUCTION The flat, all inclusive nature of long term evolution (LTE) architecture makes it ideal for integration of different mobile communication technologies. With greater commercialization of mobile communication and the surge in the number of users, the network traffic handled by LTE base station referred to as evolved NodeB (eNB) is continually increasing with time [1]. However, increment of network traffic increases the demand for limited network resources and energy consumption. References [1] and [2] state that the energy consumption by the base stations (eNB inclusive) is about 60%-80% of the total energy consumption the entire network. Therefore, [1] further described that it is difficult to modify network topology for energy saving once the eNBs have been deployed. Given this, energy planning is a parameter to be considered while deploying base stations. Three major features can be identified in handling increase in network traffic: energy consumption, limited radio resources, and equipment cost. Handling these issues while trying to maintain an acceptable level of QoS for all the users of the network with diverse traffic request is paramount to network providers. Considering minimization of energy consumption, the primary concept would be to increase the number of available radio resources to match with the growth in traffic. This approach is faced with the natural bottleneck of mobile communication: limited radio resources. In fact, LTE provides one of the most flexible bandwidth with scalability between 1.4MHz to 20MHz [3]. However, this may not be enough to handle the rapid increment of traffic due to urban migration, technology explosion and the inclusion of LTE as an access technology of internet of things (IoT). Furthermore, though LTE theoretically supports a frequency reuse factor of 1 as elucidated in [4], it cannot be achieved practically in a cellular architecture due to interference and limitations of cell edge users. Consequently, with varying frequency reuse factors (especially for cell edge users) and schemes been implemented, the number of channels available at an instantaneous scheduling time in a given eNB is reduced, thereby degrading the QoS especially under high traffic scenario [4]. Moreover, the advantages of reuse schemes are obtained in a multi-cell architecture. Conventionally, a technique that is used to manage growth in network traffic while succinctly handling the limitations of arbitrary increasing radio resources is the cell splitting technique. According to [5], “Cell splitting is the process of subdividing a congested cell into smaller cells, each with its own base station and a corresponding reduction in antenna height and transmitter power.”. In cell splitting, although the available channels for a particular eNB are reduced at an instantaneous scheduling time, the ability for multiple eNBs separated by geographical distance to use the same channels at the same time with little or no interference increases the overall capacity of the network [5], [6]. Generally, with the creation of new cells which have a smaller radius than the original cells between the existing cells, the measure of capacity increase is based on the number of channels per unit area [6]. The consequence of the cell splitting is that the frequency assignment has to be done again, which requires recalibration of neighbouring cells in order to avoid interference. It also increases the handoff rate because the cells are now smaller and a mobile is likely to cross cell boundaries more often compared with the case when the cells are bigger [5], [6]. Furthermore, 80-90% of peak energy consumed by a base station is spent on fixed equipment (i.e., equipment without direct contribution to the radio signal transmission or reception, like air conditioner, power supply, etc.) [2], [7]. The creation of new base stations due to cell splitting leads to more energy consumed by the base station. This extra-energy is primarily due to extra provisions of fixed base station equipment for the new base stations. Finally, due to the mesh topology of LTE network architecture, eNB site planning must be carefully done. Sometimes, this places a limit on how much a cell can be split. This limit is based on the availability of eNB location, access to required infrastructure and in compliance with specific country laws on base station location.

II. PROPOSED WORK To improve the challenges enumerated in section I, the development of a topology that would subtly overcome the bounds of cell splitting while maintaining and/or improving the QoS of the network is necessary. This work explored this necessity and proposes a solution that involves the installation and activation of a load-shared redundant network interface. The redundant network interface is a mini eNB with additional transceiver to complement the main eNB. This redundant interface will be placed on sleep mode in order to conserve energy until the eNB experience high traffic. When high traffic occurs, the redundant interface can be activated switching in active mode. However, when the traffic is low or returns to normal, it is switched back to sleep mode and the interface will be managed primarily by load balancing technique. Furthermore, in this work, the algorithms to implement load balancing between the eNB interfaces are presented. The proposed scenario was simulated with Network Simulator 3 (version 3.24) [8] as proposed in [9] and tested with video traffic obtained from [10].

III. CONCLUSION In conclusion, the work examined existing architectural techniques used to handle surge in network traffic in a particular area in order to maintain QoS experience by the user. Such scenarios include cell splitting and increase in network radio resources. Furthermore, we proposed a scenario that allows an extra interface to be installed at an eNB. This interface will be placed on active redundancy and managed by a load-sharing algorithm and other management techniques (such as frequency reuse). Though increase in eNB radio resources performs optimally, it is restricted by the boundary of available radio resources. The results of this scenario implementation showed that the proposed scenario conserved energy while attaining similar QoS to the cell-splitting scenario. For instance, as shown in figure 1 the average values show a 15.2% reduction of the energy consumption between the cell-splitting scenario and the load-shared redundant interface scenario. Therefore, the proposed work is a better alternative to cell splitting techniques. Nevertheless, this work has not considered cell splitting based on active antenna techniques. We have considered cell splitting based on dividing the cell and installing new base station with reduced coverage region.

ABSTRACT. Quality of Service (QoS) relies on the shaping of preferential delivery services for applications in favour of ensuring sufficient bandwidth, controlling latency and reducing packet loss. QoS can be achieved by prioritizing important broadband data traffic over the less important one. Thus, depending on the users’ needs, video, voice or data traffic take different priority based on the prevalent importance within a particular context. This prioritization might require changes in the configuration of each network entity which can be difficult in traditional network architecture. To this extent, this paper investigates the use of a QoS-based routing scheme over a Software-Defined Network (SDN). A real SDN test-bed is constructed using Raspberry Pi computers as virtual SDN switches managed by a centralized controller. It is shown that a QoS-based routing approach over SDN generates enormous control possibilities and enables automation.