Current methodologies in distributed source coding have predominantly investigated decoder-focused strategies, emphasizing the alignment and exploitation of side information. This study introduces a paradigm shift by presenting an encoder-centric algorithm that conducts proactive optimization in the frequency domain. This shift is motivated by the current deep learning models' tendency to passively extract high-frequency elements, such as contours and content in the spatial domain at the encoder side, without considering the frequency characteristics of these spatial components. Unlike current trends, the proposed scheme actively selects the essential frequency components directly in the frequency domain by introducing an adaptive self-learning filter, enabling the encoder to discern and retain critical frequency components effectively and precisely. Furthermore, we align the side information in the spatial domain before feature extraction and implement an affine transformation-based alignment strategy to utilize the side information better. By leveraging the shared frequency domain components of the image pairs, the proposed algorithm adeptly learns affine coefficients to accomplish precise spatial alignment. This dual strategy of proactive encoder optimization and decoder alignment via affine transformations is highly efficient, outperforming existing state-of-the-art methods in distributed source coding when tested across two diverse datasets by an average of 0.5 dB in PSNR.

Traditional TCP has been the dominant protocol for internet traffic after years of development in both academia and industry. However, the emergence of live video streaming applications and the increasing demand for low-latency video transmission, particularly in the context of sports and game live streaming, has posed a challenge to traditional TCP congestion control algorithms. As a representative TCP algorithm, BBR has been widely used in industry. However, BBR tends to inject more packets than the actual bottleneck in the dynamic network resulting in high latency. Because the maximum bandwidth in the past period is used to calculate the congestion window (CWND). To address this issue, we develop a recursive least squares (RLS) model to predict future bandwidth based on past bandwidth samples and update BBR's CWND periodically. To overcome the difficulty of modifying the kernel congestion control algorithm, we use extended Berkeley Packet Filter (eBPF) technology to rewrite TCP BBR and use eBPF MAP to exchange data between the kernel space and the data space. Experiments show that our algorithm can effectively reduce latency for live streaming in the dynamic network. Compared with CUBIC, our algorithm can achieve 76.9% average latency reduction with 2.3% average throughput loss and bring 42.4% average latency reduction with 1.9% average throughput loss compared with BBR.

In this paper, we propose a puncturing-based joint source channel coding scheme with a hierarchical down-sampling layer (PJSCC). The proposed hierarchical down-sampling layer fully exploits both frequency and spatial priors. Moreover, to achieve adaptive compression ratio control, PJSCC utilizes a shared puncturing table as a global prior shared between the sender and receiver. This puncturing table plays a vital role in selectively pruning or padding symbols, and the adaptation of the compression ratio is achieved through the manual configuration of hyperparameters to adjust the puncturing rate. Experimental results show that the proposed scheme achieves superior reconstruction performance across several classic datasets.

Current point cloud compression methods based on deep learning cannot guarantee that the reconstructed points are constrained to the surface, resulting in low reconstruction quality at low bitrates. Hence, this paper proposes an efficient deep learning-based point cloud geometry compression algorithm. Specifically, by introducing a two-dimensional plane at the decoder, the reconstructed local patch is constrained within a manifold, preserving sufficient surface features. This strategy ensures the decoder can reconstruct high-quality point clouds even at low bitrates. Moreover, we use the anchor features obtained by the neural network to compress the local features at the encoder. The experimental results show that, under the condition of the same restoration quality, the proposed method improves the point-to-plane PSNR by more than 2dB compared to the state-of-the-art methods. The code is available at https://github.com/zbaoye/SurfPCC.

Mobile-edge computing (MEC) technology offers computing resources for mobile devices to conduct computationally heavy activities by putting servers at the wireless mobile network’s edge. This mitigates the scarcity of computing resources in mobile devices and enhances the intelligence of the Internet of Things (IoT), which is a crucial technology for achieving industrial digitalization. Considering the time-varying channel as well as the time-varying available computing resources of MEC servers, this article formulates a hybrid optimization problem that combines task offload and resource allocation. The goal is to minimize MEC servers’ overall power consumption. Since the channel state information (CSI) stored in the MEC system is not real time, we propose a reinforcement learning (RL) algorithm for predicting current CSI from historical CSI and obtain the optimal strategy for task offloading. On the other hand, convex optimization methods are used to accomplish the dynamic resource allocation strategy. In addition, an approach based on deep RL (DRL) is put forward to overcome the dimensionality curse in RL algorithms. The simulation experiments illustrate that the proposed algorithms outperform the nonpredictive schemes by a large margin, and their performance is close to that of the optimum scheme, which utilizes simultaneous CSI.

The progress of the deep neural network and visual sensors promote the facial landmark detection. However, faces are easily collected from Internet of Things (IoT) devices worldwide but are hard-labeled facial landmarks in consistent style. Though many weak supervision (WS) algorithms and theories have been proposed to handle the labeling problem, most of them are tailored for classification tasks and fail in regression tasks, especially in facial landmark detection. To tackle this WS regression task, first, we propose a regressive pseudo-labeling method by analyzing the cluster assumption of facial landmark detection, where overlaps are reduced and interval areas are increased among clusters of facial parts for unlabeled faces. Moreover, auxiliary information and domain loss are utilized to adapt model to samples with different styles. Second, we design a generator–regressor network to first estimate facial boundary attentions and then to locate facial landmarks. However, when generating pseudo labels based on predictive models automatically, there are two major issues. One is that the results of the multistage network highly depend on the former-stage accuracy, and another is that jointing different annotation styles always produces ambiguity feature representation. Thus, we propose two ideas. During training, the generator and regressor are decoupled to alleviate the inner dependence of the multistage network, and the heatmap discriminator is introduced to improve the quality of the predicted facial boundary. We design a transformer structure to fuse face image and boundary attentions, so that it can further complement useful features. Based on these ideas, our methods can automatically annotate accurate pseudo facial landmarks for unlabeled faces. Extensive experiments show that our model achieves good performance on different benchmark data sets, both in accuracy and efficiency.

This paper provides a method to build a deep learning image coding system based on inverse problem, choosing a suitable measurement operator to reduce the amount of information transmitted at the sender, and reconstructing the original image by tackling the inverse problem at the receiver. Unlike most compressed sensing (CS) methods, the proposed coding scheme does not rely on sparsity but uses the structural priors of the generative adversarial networks (GAN) to solve the inverse problem. The proposed model trains the GAN to learn a mapping from the latent space to the sample space formed by correlated images on the cloud. Then the measurements are used to localize the optimal latent variable in the representation space which corresponding to the original image in the sample space. The proposed method encodes and transmits the measurements instead of the original image, which greatly reduces the cost of transmission while ensuring the quality of the reconstructed the image at high compression ratios. To the best of our knowledge, this is the first time to introduce the GAN-based inverse problem in the field of the deep image coding area. The experimental results show that the visual quality of the images generated by the proposed scheme is better than the traditional encoding scheme JPEG2000. Especially in the case of extremely high compression ratios, the proposed scheme can still maintain good performance.

Cloud VR (Virtual Reality) is a VR scheme based on cloud computing, which can reduce the computing burden on terminal equipment. It uses edge computing for adaptive streaming to minimize the response latency and bandwidth consumption. However, traditional adaptive streaming approaches based on preprocessing have some limitations. In this paper, we proposed a novel adaptive Cloud VR system with real-time processing. We design and implement a GPU acceleration scheme to perform efficient projection and coding, which makes the computing latency acceptable. The scheme is further extended to pipeline execution with simple orientation prediction to support higher frame rate. The real-time processing not only reduces the storage size, but also eliminates the drawbacks of pre-generating limited video versions. According to the experimental results, our scheme can provide a VR stream matching user viewport more precisely. Through optimized GPU algorithm, we shorten the processing time to 1/10 of the original. Compared to classic pyramid projection scheme, our system effectively reduces the average orientation deviation by 90.23%, thus provides more robust service.

Joint pushing and caching are commonly considered an effective way to adapt to tidal effects in networks. However, the problem of how to precisely predict users' future requests and push or cache the proper content remains to be solved. In this paper, we investigate a joint pushing and caching policy in a general mobile edge computing (MEC) network with multiuser and multicast data. We formulate the joint pushing and caching problem as an infinite-horizon average-cost Markov decision process (MDP). Our aim is not only to maximize bandwidth utilization but also to decrease the total quantity of data transmitted. Then, a joint pushing and caching policy based on hierarchical reinforcement learning (HRL) is proposed, which considers both long-term file popularity and short-term temporal correlations of user requests to fully utilize bandwidth. To address the curse of dimensionality, we apply a divide-and-conquer strategy to decompose the joint base station and user cache optimization problem into two subproblems: the user cache optimization subproblem and the base station cache optimization subproblem. We apply value function approximation Q-learning and a deep Q-network (DQN) to solve these two subproblems. Furthermore, we provide some insights into the design of deep reinforcement learning in network caching. The simulation results show that the proposed policy can learn content popularity very well and predict users' future demands precisely. Our approach outperforms existing schemes on various parameters including the base station cache size, the number of users and the total number of files in multiple scenarios.

A generalized hybrid digital-analog (HDA) framework with the combination of orthogonal and nonorthogonal multiplexing is proposed, which can strike a balance between interference and resource for Internet of Things application. The optimal resource allocation for the proposed scheme is formalized as a 3-D mixed integer programming problem, which is a function of digital bandwidth, orthogonal power, and nonorthogonal power of analog signal. With divide and conquer strategy, we first search the space of digital bandwidth, which is constructed by the possible number of subcarriers in orthogonal frequency division multiplex system, then the optimization problem is reduced to a 2-D continuous optimization problem. We further decompose it into two 1-D continuous optimization problems, and prove they are convex 1-D functions unconditionally or conditionally, respectively. With their convexity, the 2-D optimization problem can be solved with iterative gradient descent algorithm. We design a resource allocation algorithm to solve the optimization problem in practical system. Our experimental results show that the proposed algorithm outperforms nonorthogonal multiplexing HDA by 1-3 dB in terms of peak signal to noise ratio.

How to strike a balance between diversity gain and spatial multiplexing gain in a soft video delivery system is an open problem. Due to the power limit, it is especially important to achieve the optimal balance during video transmission in the Internet of Things. In this paper, taking the multisimilarity feature of a soft video system into account, we design an adaptive multiple-input, multiple-output (MIMO) receiver that can utilize nearly optimally either diversity gain or multiplexing gain. At the transmitter, we arrange multisimilar video data according to the space-time coding style and transmit them through multiple antennas. At the receiver, we propose using two decoders, i.e., the multisimilar space-time block coding (Ms-STBC) decoder and the soft MIMO decoder. The decoder to be chosen is determined by the predicted performance gain. We show that the proposed Ms-STBC transmission scheme can be considered a joint source-channel design, which is proposed for a soft video multiantenna delivery system. The relationship between intracodewords similarity and the channel signal-to-noise ratio (SNR) gain is derived. The experimental results demonstrate that the proposed designs can achieve a significant improvement over either individual soft MIMO multiplexing decoding or individual space-time block coding decoding in terms of peak SNR (PSNR) under the condition of a time-varying channel and a wide range of SNR. We can obtain at most 4-dB PSNR gain compared with the soft MIMO system.

Hybrid digital-analog (HDA) video transmission is a new cross-layer design, which can be widely used in Internet of Things. The key problem of HDA video transmission is to find the optimal resource allocation between the digital and analog part. This paper presents a new general resource allocation algorithm for superposition coding-based HDA system. On one hand, in order to achieve successful decoding in digital part, the bitrate is controlled by the quantization parameter (QP), and the channel coding rate and modulation order are determined by the signal to interference noise power ratio, in which analog part is considered as the interference. On the other hand, the overall video quality is directly determined by the mean square error of analog part, which depends jointly on the data variance of the analog part, the power allocated to the analog part and the channel noise power. We propose a prediction model to describe how the data variance of the analog part changes with the QP in the digital part. Based on the proposed model, the power allocation of two parts can be quantitatively connected to form an optimization problem. We prove the convexity of the resource allocation problem and the gradient descent method is utilized in system implementation. With extensive simulations, the proposed algorithm is validated, achieving 1.4-5.3 dB gain over the conventional digital system, and 6.2-7.4 dB gain over pseudo-analog system in peak signal-to-noise ratio.

This paper presents a network slice design for ultra high definition (UHD) video broadcast/multicast to achieve higher network efficiency and improved quality of experience (QoE). The proposed network slice design consists of a rateless source compression scheme and an analog-coded SoftCast scheme. The rateless Spinal code is adopted to compress the video source at content server and the compressed source is transmitted from content server across wireless core network to the base station. Ana prioriinformation-assisted Spinal decoder is designed to utilize the sparsity of bit planes for compression. In the analog-coded SoftCast scheme, we design a new chaotic function-based analog code with negligible power penalty for the generalized Gaussian-distributed source in SoftCast because the existing chaotic functions designed for uniformly distributed sources suffer from serious power penalty in SoftCast. We also design a maximuma posterioriprobability decoding algorithm for the proposed analog code in order to exploit the statistics of video source asa prioriinformation to improve the performance. The experimental results show that the proposed rateless code-based compression scheme achieves efficient compression and approaches the bound of binary erasure channel. In particular, the 1/2 analog-coded SoftCast has almost 2 dB gain over conventional SoftCast with two repetitions, and the 1/3 analog-coded SoftCast has almost 3 dB gain over conventional SoftCast with three repetitions. The system simulations for the broadcast system show higher network capacity and improved QoE in the proposed UHD slice, because the reconstructed video quality of each user is commensurate with its channel condition.