Workshops and Tutorials
3rd Workshop on Intelligent Computing and Caching at the Network Edge
Session FWS6-S1
Caching in Heterogenous Wireless Network
Conference
9:00 AM
—
10:30 AM KST
Local
May 24 Sun, 8:00 PM
—
9:30 PM EDT
Keynote: Scheduling of Multiple Tasks among Multiple Helpers in Fog/Edge Network
Yang Yang (ShanghaiTech University, China)
1
This talk does not have an abstract.
Wind-aware Content Caching for Smart Farm
Seng-Kyoun Jo (ETRI, Korea (South)); Max Muehlhaeuser (Technical University Darmstadt, Germany); Se-han Kim (ETRI(Electronics and Telecommunications Research Institude), Korea (South))
0
In this paper, we propose a new sustainable approach for content-centric networking (CCN) where content caching performs towards windy-aware content delivery. We design a caching strategy, where we first define the windiness of nodes, a quantitative metric for measuring the available power generation from wind turbines, based on which we identify corresponding windy paths and encourage content to aggregate on windy paths powered by more wind energy. We validate our approach with a variety of simulations using real smart farm testbed topology and its meteorological datasets from Rep. of Korea, and the results show that applying the proposed wind-aware caching scheme achieves significant gains by reducing existing fossil-fuel energy.
Joint Power and Channel Allocation Based on Mobility and Interest Aware D2D Cache in HetNets
Xinpeng Lyu, Ying Wang, Zhendong Li and Man Liu (Beijing University of Posts and Telecommunications, China)
0
Device-to-device (D2D) communication has become a key technology to solve the outbreak of short video traffic in the network. There are many users with different interests in the Internet, and different users request diverse videos when they need them. At the same time, users are constantly moving, which will affect the distance of D2D communication. Therefore, the user's mobility and user's interests should be taken into account when designing the cache scheme. Based on user mobility and user interest, this paper formulates a problem of maximizing the energy spectral efficiency of joint user connection selection, bandwidth allocation and power allocation based on mobility and interest aware D2D cache in HetNets. This problem is a fractional form and not a convex problem, so it is hard to be solved directly. We decouple the problem into three subproblems, and transform the problem into a subtraction form by Dinkelbach algorithm. We propose an iterative algorithm to obtain the optimized result. Simulation results show that the algorithm proposed effectively improves the energy spectral efficiency of the network.
Utility Maximization for Cache-Aided Ultra-Dense Relay Networks: A Matching Perspective
Yuqin Liu and Feng Ke (South China University of Technology, China); Hui Song (South China Normal University, China)
0
In order to deal with the surge of data traffic, the caches of the nodes in a wireless network can be utilized in efficient ways. This paper investigates the utility optimization for ultra-dense relay networks (UDRNs) by a preference-aware caching and stable matching scheme. First, we divide the nodes, which may act as a user or a relay, into multiple clusters, and perform optimization strategy within the clusters. Second, the file caching scheme is designed by taking user preference and file popularity into account. Third, according to the performances of the file caching scheme, the optimal power bought from the relay to maximize the utility of the user is derived by an iterative algorithm, and the mutual preference matrices between the users and relays by the maximum utility criterion is established. Then, we propose a stable matching strategy by exploring the Gale-Shapley algorithm. Simulation results demonstrate that the proposed algorithm can bring significant performance improvements compared with the conventional algorithms.
Session Chair
Sheng Zhou (Tsinghua University, P.R. China)
Session FWS6-S2
Coded Caching at Network Edge
Conference
10:45 AM
—
12:15 PM KST
Local
May 24 Sun, 9:45 PM
—
11:15 PM EDT
Optimized Coded Prefetching Scheme in Hierarchical Cache-Enabled Networks
Yan Tan (Harbin Institute of Technology(shenzhen), China); Ye Wang (Harbin Institute of Technology (Shenzhen), China); Shushi Gu (Harbin Institute of Technology, Shenzhen, China); Xianfan Sun (Harbin Institute of Technology(shenzhen), China); Qinyu Zhang (Shenzhen Graduate School, Harbin Institute of Technology, China); Wei Xiang (James Cook University, Australia)
0
Caching popular content at small base stations (SBSs) of a wireless edge network is a good choice to reduce user request latency and backhaul load. However, an effective coded caching scheme to solve the challenges in hierarchical cache-enabled networks (HCENs), including the limited SBS
cache capacity, rich content amount, especially the deployment of coded prefeching, has not yet been fully studied. In this paper, we propose a caching scheme based on coded prefetching and coded transmission in HCENs. In particular, considering the impact of preference of users, the SBSs cache capacity, and the number of coded linear combinations of content, we design a
caching probability matrix, where content is stored at each SBS by linear combinations with caching probability. Furthermore, the proposed algorithm including prefetching and transmission is analyzed, the expression of the system average delay is derived, and an optimization problem of minimizing the average delay is established to obtain an optimal cache probability matrix.
Finally, the simulation results verify the effectiveness of our coded prefetching caching scheme, which can achieve lower average delay than the uniform random caching scheme and most popular caching scheme in HCENs.
cache capacity, rich content amount, especially the deployment of coded prefeching, has not yet been fully studied. In this paper, we propose a caching scheme based on coded prefetching and coded transmission in HCENs. In particular, considering the impact of preference of users, the SBSs cache capacity, and the number of coded linear combinations of content, we design a
caching probability matrix, where content is stored at each SBS by linear combinations with caching probability. Furthermore, the proposed algorithm including prefetching and transmission is analyzed, the expression of the system average delay is derived, and an optimization problem of minimizing the average delay is established to obtain an optimal cache probability matrix.
Finally, the simulation results verify the effectiveness of our coded prefetching caching scheme, which can achieve lower average delay than the uniform random caching scheme and most popular caching scheme in HCENs.
Online Caching and Coding at the WiFi Edge: Gains and Tradeoffs
Lalhruaizela Chhangte (IITB-Monash Research Academy, India); Emanuele Viterbo (Monash University, Australia); D. Manjunath (IIT Bombay, India); Nikhil Karamchandani (Indian Institute of Technology Bombay, India)
2
Video content delivery at the wireless edge continues to be challenged by insufficient bandwidth and highly dynamic user behavior which affects both effective throughput and latency. Caching at the network edge and coded transmissions have been found to improve user performance of video content delivery. The cache at the wireless edge stations (BSs, APs) and at the users' end devices can be populated by pre-caching content or by using online caching policies. In this paper, we propose a system where content is cached at the user of a WiFi network via online caching policies, and coded delivery is employed by the WiFi AP to deliver the requested content to the user population. The content of the cache at the user serves as side information for index coding. We also propose the LFU-Index cache replacement policy at the user that demonstrably improves index coding opportunities at the WiFi AP for the proposed system. Through an extensive simulation study, we determine the gains achieved by caching and index by coding. Next, we analyze the tradeoffs between them in terms of data transmitted, latency, and throughput for different content request behaviors from the users. We also show that the proposed cache replacement policy performs better than traditional cache replacement policies like LRU and LFU.
Coded Caching with Heterogeneous User Groups
Jingliang He (Sun Yat-sen Unviersity, China); Congduan Li (Sun Yat-sen University, China); Linqi Song (City University of Hong Kong, Hong Kong)
1
Coded caching is a promising technique to reduce the backbone load in content requesting and delivery networks. In conventional setup, users are assumed to be homogeneous such that all users cache the files in a same manner, i.e., for any two users, they cache the files according to the same distribution, which could be a uniform caching distribution or a nonuniform one. However, in the perspective of video service providers, it's necessary to offer better service for paying subscribers or VIP users, compared to normal users, to decrease the network load. In this paper, a more general and practical case with heterogeneous users is investigated. Users are classified as two groups, say VIP and non-VIP group. We would like to see how coded caching performs to ensure the users in VIP group have a better experience than those in non-VIP group. These two user groups are considered with nonuniform and uniform caching distributions, respectively. A heterogeneous caching and delivery scheme is proposed and the associate upper bound on the achievable rate is obtained with the robustness of fractions of the two groups. Simulation results confirm the robustness of our theoretical results and show that the VIP group could have better experience than the non-VIP group.
User Movements Aware Coded Caching in Small-Cell Networks
Guangyu Zhu, Caili Guo and Tiankui Zhang (Beijing University of Posts and Telecommunications, China); Qianqian Yang (Imperial College London, United Kingdom (Great Britain))
0
In coded caching, users cache pieces of files under a specific arrangement so that the server can satisfy their requests simultaneously in the broadcast scenario via eXclusive OR (XOR) operation and therefore reduce the amount of transmission data. As a character of wireless communication networks, user movements, however, make the caching arrangement fruitless, which results in the disappearance of XOR opportunities. In this paper, we propose a coded caching scheme to address this problem caused by user movements. The proposed scheme divides users into groups according to their caching patterns in order that XOR coded caching messages can be reorganized when user movements happen. Transmission data volume is driven to measure the performance of the proposed scheme. Numerical results show that the proposed scheme achieves improvement on traffic offloading.
Session Chair
Zhiyuan Jiang (Shanghai University, P.R. China)
Session FWS6-S3
Intellignet Edge Computing
Conference
2:00 PM
—
3:30 PM KST
Local
May 25 Mon, 1:00 AM
—
2:30 AM EDT
Keynote: Big Data Analysis and Cross-layer Optimization for Communications, Caching and Computing Networks
Zhu Han (Houston University, USA)
1
This talk does not have an abstract.
Mobility-Aware Cooperative Task Offloading and Resource Allocation in Vehicular Edge Computing
Yifan Zhang, Xiaoqi Qin and Xianxin Song (Beijing University of Posts and Telecommunications, China)
2
Vehicular edge computing (VEC) is considered as a promising technology to support emerging applications in vehicular networks by leveraging the ubiquitous computing resource at network edge. To leverage the excessive computing power on road side units (RSUs) and neighboring vehicles, effective collaboration among the vehicles and between vehicles and RSUs is required. The high mobility of vehicles and the ad hoc nature of networking introduces difficulties in effective task scheduling and reliable result feedback. In this paper, we study a joint task offloading and resource allocation problem in vehicular networks with the aim of optimizing the system utility related to delay and cost of computing and communication services. The problem formulation involves joint consideration of mobility of vehicles, task offloading decision, computing resource allocation, and reliable result feedback. The formulated problem falls in the form of a mixed-integer nonlinear program (MINLP). To solve the problem efficiently, we propose a light-weight matching based task offloading and resource allocation algorithm. Through simulation results, we show that the performance of our proposed scheme is competitive when compared with existing strategies.
Joint User Association and Value-Aware Computation Offloading for MEC-Enabled Networks
Huiwen Zhang and Wenpeng Jing (Beijing University of Posts and Telecommunications, China); Zhaoming Lu (BUPT, China); Xiangming Wen and Jingyi Zhang (Beijing University of Posts and Telecommunications, China)
2
Computation offloading in mobile edge computing (MEC) plays an important role in mitigating the burden on users. However, there remain some obstacles hindering the full use of the benefits of computation offloading in MEC-enabled networks. Besides, the intensive deployment of access points (APs) integrated with MEC servers (MESs) leads to a user association dilemma, which makes the offloading decision and resource allocation more challenging. In this paper, we investigate the optimization problem of user association, offloading decision, and resource allocation for MEC-enabled networks. Different from the existing works that always make utility optimization from users' perspective, an optimization problem is formulated in this paper, which aims to improve the potential benefits from the operator's perspective under the constraints of users' quality of service (QoS). Moreover, we present a value-aware scheme to classify tasks so that the operator can make reasonable pricing for different performance requirements of tasks. Due to the non-convex property of the above problem, we propose an iterative framework to decouple the original problem into two sub-problems, namely a user association problem and a joint offloading decision and resource allocation problem. These two problems are solved by matching theory and block coordinate reduction (BCD), respectively. A User Association based on Matching Theory (UAMT) algorithm and a Joint Optimization of Offloading Decision and Resource Allocation based on BCD (JOODRA-BCD) algorithm are designed correspondingly. The simulation results verify that the proposed algorithm has a fast convergence property and outperforms the benchmark algorithms in terms of the utility of the operator.
Intelligent Deployment of Dedicated Servers: Rebalancing the Computing Resource in IoT
Yiwen Wu (University of Electronic Science and Technology of China, Chengdu, China); Yilin Wang, Yunkai Wei and Supeng Leng (University of Electronic Science and Technology of China, China)
1
The applications in Internet of Things (IoT), which are increasingly diversified and complicated, are burdening ever-more higher computing requirements on the IoT devices. However, these devices are mostly lightweight nodes with limited computing resources. Some access points, such as the sinks or gateways, can alleviate the computing burden of the lightweight nodes with the assistance of computation offloading, e.g., edge computing or fog computing. Whereas, this is mostly limited by the location, computing power, and quantity of such access points. An emerging method to solve this issue is to deploy appropriate servers to provide computing service in the network, since it is impossible to upgrade all lightweight nodes into resource-rich nodes. In this paper, we propose an intelligent deployment of the computing servers (IDCS) based on a genetic algorithm, which can achieve optimized latency and load balance in deploying the computing servers to form hybrid computing service providers in the IoT network. Extensive simulations show that IDCS can outperforms traditional schemes considerably in aspects of both latency and load balance.
Session Chair
Zheng Chang (University of Jyväskylä, Finland)
Session FWS6-S4
Enabling the Network Edge Intelligence
Conference
3:45 PM
—
5:15 PM KST
Local
May 25 Mon, 2:45 AM
—
4:15 AM EDT
Tag Selection for Backscatter Communication in Classified Wireless Body Area Networks
Zhuang Ling and Fengye Hu (Jilin University, China); Li Dong (Macau University of Science and Technology, Macao); Zhu Han (University of Houston, USA)
0
In this paper, a tag selection protocol for Backscatter Communication (BackCom) in Classified Wireless Body Area Networks (CWBANs) with wireless energy transfer (WET) is proposed, which includes an access point (AP) and multiple tags. To solve the problem of tags overuse and channel mismatch in CWBAN, we consider a training phase and a BackCom transmission phase in each transmission block. In the training phase, the AP firstly transmits training RF signals with WET to the tags. The tags with maximum energy and the most important priority are selected from all tags in CWBAN, which is the subset of CWBAN and named as subclassification WBAN (SCWBAN). In the BackCom phase, the tag nodes in SCWBAN firstly receive command information from the AP, and then backscatter physiological information to the AP. The tag nodes with the best channel state information are selected from the SCWBAN. From the two phases, both energy of the tag nodes and channel state information are considered for multiple tags selection. According the characteristics of CWBAN, the closed form of the analytical outage probability is derived, from which the insight on how the channel and system parameters affect the outage performance is gained. Finally, simulation results are provided to corroborate our analytical results.
VNF Placement and Resource Allocation in SDN/NFV-enabled MEC Networks
Nahida Kiran, Xuanlin Liu, Sihua Wang and Changchuan Yin (Beijing University of Posts and Telecommunications, China)
0
Network function virtualization (NFV), software defined networks (SDNs), and mobile edge computing (MEC) are emerging as core technologies to satisfy increasing number of users' demands in 5G and beyond wireless networks. SDN provides clean separation of the control plane from the data plane while NFV enables the flexible and on-the-fly creation and placement of virtual network functions (VNFs) and are able to be executed within the various locations of a distributed system and, in our case, in the NFV-enabled MEC nodes. VNF placement and resource allocation (VNFPRA) problem is considered in this paper which involves placing VNFs optimally in distributed NFV-enabled MEC nodes and assigning MEC resources efficiently to these VNFs to satisfy users' requests in the network. Current solutions to this problem are slow and cannot handle real-time requests. To this end, we propose an SDN-NFV infrastructure to tackle the VNFPRA problem in wireless MEC networks. Our aim is to minimize the overall placement and resource cost. Two algorithms are proposed: (i) an optimal solution formulated as a MIP problem (ii) a genetic based heuristic solution. The superior performance of the proposed solution is confirmed in comparison with two existing algorithms such as Random-Fit Placement Algorithm (RFPA) and First-Fit Placement Algorithm (FFPA). The results demonstrate that a coordinated placement of VNFs in SDN/NFV enabled MEC networks can satisfy the objective of overall reduced cost. Simulation results also reveal that the proposed scheme approximates well with our optimal solution returned by gurobi and also achieves reduction on overall cost.
Analysis of Group Distribution and Content Concentration for Packet Allocation in D2D Communication
Kuan Wu, Lei Zhao, Ming Jiang and Xiaojing Huang (Sun Yat-sen University, China)
0
Among the many popular issues that have been identified in the area of device-to-device (D2D) based packet caching technologies, a new aspect related to the competitions between various D2D user equipment (UE) groups, is not fully studied. Hence, in this paper, we investigate a few key perspectives of the competitions among UE groups with respect to the packet allocation strategy. We first analyze the impact from diverse group proportions on the system throughput and the fairness of packet allocations. Particularly, the new concept of group separation index (GSI) is introduced to help reflect the fairness of system-level packet allocations. Then, we evaluate how the concentration levels of diverse packet requests may affect the packet allocation strategy. Finally, simulation results validate our analysis and show the potential of the proposed analysis method. The new approach can provide important design hints for improving existing D2D-based packet caching schemes.
End-to-End Delay Analysis in mmWave UAV-assisted Wireless Caching Networks
Kai-Min Liao, Guan-Yi Chen, Yu-Jia Chen and Yung-Fang Chen (National Central University, Taiwan)
0
In this paper, we analyze the end-to-end delay bound of the uplink transmission in an unmanned aerial vehicle (UAV)-assisted wireless network, where the UAV acts as the edge caching node to enhance the quality of service for content providers. Due to the emerging wireless connected intelligent devices, it is expected that the uplink traffic will grow dramatically. Recently, cache-enabled UAV base stations (UAV-BSs) are proposed to relieve the massive upload contents from neighboring devices. Meanwhile, since the storage tasks are executed close to the devices, the network latency can be reduced, which is a promising way to provide delay-sensitive services. To understand the performance impact of UAV-based wireless caching, we adopt network calculus to analyze the upper bound of uplink latency. The proposed model takes into account the transmission failure probability caused by the mobility of UAVs. Our numerical results confirm the performance gain of using cache-enabled UAV-BSs in wireless mobile networks.
Session Chair
Jie Gong (Sun Yat-sen University, P.R. China)