1-4hit |
Rongqi ZHANG Chunyun PAN Yafei WANG Yuanyuan YAO Xuehua LI
With maturation of 5G technology in recent years, multimedia services such as live video streaming and online games on the Internet have flourished. These multimedia services frequently require low latency, which pose a significant challenge to compute the high latency requirements multimedia tasks. Mobile edge computing (MEC), is considered a key technology solution to address the above challenges. It offloads computation-intensive tasks to edge servers by sinking mobile nodes, which reduces task execution latency and relieves computing pressure on multimedia devices. In order to use MEC paradigm reasonably and efficiently, resource allocation has become a new challenge. In this paper, we focus on the multimedia tasks which need to be uploaded and processed in the network. We set the optimization problem with the goal of minimizing the latency and energy consumption required to perform tasks in multimedia devices. To solve the complex and non-convex problem, we formulate the optimization problem as a distributed deep reinforcement learning (DRL) problem and propose a federated Dueling deep Q-network (DDQN) based multimedia task offloading and resource allocation algorithm (FDRL-DDQN). In the algorithm, DRL is trained on the local device, while federated learning (FL) is responsible for aggregating and updating the parameters from the trained local models. Further, in order to solve the not identically and independently distributed (non-IID) data problem of multimedia devices, we develop a method for selecting participating federated devices. The simulation results show that the FDRL-DDQN algorithm can reduce the total cost by 31.3% compared to the DQN algorithm when the task data is 1000 kbit, and the maximum reduction can be 35.3% compared to the traditional baseline algorithm.
Video applications such as video conferencing among multiple users and video surveillance systems require multiple video connections and QoS guarantee. These days the video systems equipped with IEEE 802.11 LAN interfaces allows a broadband wireless access to the Internet at a reasonable price. However, according to the current IEEE 802.11e HCCA standard, if more than two video sessions are to be established simultaneously, some of them must share the TXOP because the available number of TSIDs for video transmission is restricted to two. In order to resolve this problem, we devise a scheme which can establish up to 13 video sessions by slightly modifying the frame structure while maintaining the compatibility with the current standard. Our scheme is implemented on the NCTUns 4.0 network simulator, and evaluated numerically in terms of throughput, delay, and PSNR. Also real video clips are used as input to our simulation. The results showed that our scheme guarantees the transmission bandwidth requested by each video session.
Jong-Ok KIM Toshiaki YAMAMOTO Akira YAMAGUCHI Sadao OBANA
To meet the bandwidth requirements of multimedia services, multipath transmission is a promising solution. In this paper, we consider multi-access networks, where WiMAX and WiFi links are set up at the same time. Multipath transmission suffers from the intrinsic problem of out-of-order packet delivery. This has an adverse impact on TCP and even UDP-based delay sensitive applications. However, multimedia streaming services allow some tolerance to transmission delay. Motivated by this observation, we investigate how to split multimedia flows over heterogeneous links. Wireless link capacity varies widely over time due to dynamic radio conditions. The capacity variations should be promptly reflected in traffic splitting in order to accomplish an equal load-balance. A practical prototype system has been implemented. We have performed extensive measurements from a prototype system. Through practical experimental results, we could verify two major research goals. One is that multimedia splitting can improve the overall network performance (e.g., the permitted multimedia sessions or the aggregated bandwidth) while still keeping an acceptable media quality. The other is an adaptation capability to varying link quality. It has been widely investigated under various radio conditions and different monitoring intervals. It is shown that the adaptive technique is effective under dynamic radio environments.
Sang Hyuk KANG Min Young CHUNG Bara KIM
In this letter, we propose a video traffic model based on a class of stochastic processes, which we call truncated GeoY/G/∞ input processes. Group of picture (GOP) size traces are modeled by truncated GeoY/G/∞ input process with gamma-distributed batch sizes Y and Weibull-like autocorrelation function. With full-length MPEG-4 video traces in QCIF, we run simulations to show that our proposed model estimates packet loss ratios at various traffic loads more accurately than existing modeling methods.