Wei Su

Wei Su


Research Focus & Metrics

Academic Career Length

17Years (2010-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
3 September 2021 in IEEE Transactions on Network and Service Management
Verified
Journal Article
Closed access
GHGuobiao He
Guobiao He
WSWei Su
Wei Su
SGShuai Gao
Shuai Gao
NLNingchun Liu
Ningchun Liu
SSajal K. Das
Sajal K. Das
Multi-domain networking slice orchestration is an essential technology for the programmable and cloud-native 5G network. However, existing research solutions are either based on the impractical assumption that operators will reveal all the private network information or time-consuming secure multi-party computation which is only applicable to limited computation scenarios. To provide agile and privacy-preserving end-to-end network slice orchestration services, this paper proposes NetChain, a multi-domain network slice orchestration architecture based on blockchain and trusted execution environment. Correspondingly, we design a novel consensus algorithm CoNet to ensure the strong security, scalability, and information consistency of NetChain. In addition, a bilateral evaluation mechanism based on game theory is proposed to guarantee fairness and Quality of Experience by suppressing the malicious behaviors during multi-domain network slice orchestration. Finally, the prototype of NetChain is implemented and evaluated on the Microsoft Azure Cloud with confidential computing. Experiment results show that NetChain has good performance and security under the premise of privacy-preserving.

28 January 2022 in IEEE Transactions on Network and Service Management
Verified
Journal Article
Closed access
LMLisheng Ma
Lisheng Ma
WSWei Su
Wei Su
BWBin Wu
Bin Wu
BYBin Yang
Bin Yang
Xiaohong JiangXiaohong Jiang
Xiaohong Jiang
As important network infrastructures to support data storage and service delivery for worldwide users, cloud data centers are facing great threaten by frequent disasters around the world and thus the survivability of cloud data centers becomes a critical issue. Since both data and service evacuations are desired at the same time under a real disaster scenario, this paper studies a joint design of them to fight against disasters. We consider a disaster that can present an early warning time before it really affects cloud data centers, and by exploiting the intrinsic interplay between data and service evacuations and efficiently utilizing the early warning time we propose a joint data and service evacuation scheme for emergency protection. We first formulate the joint design as two optimal Integer Linear Program (ILP) models. Notice that the protection process is highly time-sensitive due to the early warning time constraint, two time-efficient heuristics are then designed by carefully selecting evacuated services and candidate evacuation nodes to achieve a better sharing of network resources between data backup and service migration. Extensive numerical results demonstrate the efficiency of the proposed scheme on improving survivability of data and services in cloud data centers. With a set of given resource and early warning time constraints, this work can guide data center operators to achieve a tradeoff between data backup and service migration.

1 July 2022 in IEEE Transactions on Communications
Verified
Journal Article
Closed access
JHJi He
Ji He
JLJia Liu
Jia Liu
WSWei Su
Wei Su
YSYulong Shen
Yulong Shen
Xiaohong JiangXiaohong Jiang
Xiaohong Jiang
NSNorio Shiratori
Norio Shiratori
This paper explores the joint secrecy and delay guarantees based on opportunistic jamming and link selection in a wireless relay system consisting of a source, a destination, multiple buffer-aided relays and a passive eavesdropper wiretapping over both hops. Based on the information of link state and buffer status, we design a novel transmission scheme based on link selection and jammer selection, which dynamically grants transmission links different priorities for packet delivery, such that the constraints on both secrecy outage probability and packet delay are jointly satisfied. To understand the performance of the new scheme, we then apply the bitmap technique and Markov chain theory to develop a complete theoretical framework for the modelling of three fundamental metrics, namely reliability outage probability, packet discarding probability and secrecy/delay constrained throughput (SDT). Finally, we provide extensive simulation and numerical results to validate our theoretical modelling, as well as to demonstrate that the proposed scheme is superior to the benchmarks in terms of SDT.

29 November 2023 in IEEE Transactions on Vehicular Technology
Verified
Journal Article
Open access
GHGaofeng Hong
Gaofeng Hong
BYBin Yang
Bin Yang
WSWei Su
Wei Su
HLHaoru Li
Haoru Li
ZHZekai Huang
Zekai Huang
TTTarik Taleb
Tarik Taleb
Caching the high definition map (HDM) on the edge network can significantly alleviate energy consumption of the roadside sensors frequently conducting the operators of the traffic content updating and transmission, and such operators have also an important impact on the freshness of the received content at each vehicle. This paper aims to minimize the energy consumption of the roadside sensors and satisfy the requirement of vehicles for the HDM content freshness by jointly scheduling the edge content updating and the downlink transmission resource allocation of the Road Side Unit (RSU). To this end, we propose a deep reinforcement learning based algorithm, namely the prioritized double deep R-Learning Networking (PRD-DRN). Under this PRD-DRN algorithm, the content update and transmission resource allocation are modeled as a Markov Decision Process (MDP). We take full advantage of deep R-learning and prioritized experience sampling to obtain the optimal decision, which achieves the minimization of the long-term average cost related to the content freshness and energy consumption. Extensive simulation results are conducted to verify the effectiveness of our proposed PRD-DRN algorithm, and also to illustrate the advantage of our algorithm on improving the content freshness and energy consumption compared with the baseline policies.

19 March 2021 in IEEE Networking Letters
Verified
Journal Article
Closed access
LTLizhuang Tan
Lizhuang Tan
WSWei Su
Wei Su
JMJingying Miao
Jingying Miao
WZWei Zhang
Wei Zhang
In-band network telemetry (INT) is a new network measurement technology that uses business packet to collect network information hop by hop. Since business packets may be lost due to various reasons, INT telemetry information will inevitably be lost. Unfortunately, telemetry system is not aware of this loss. In this letter, we present FindINT, an INT packet coding scheme with two marking strategies, to measure per-flow packet loss rate and location. FindINT is inspired by alternative marking method, which is triggered by loss event instead of polling.Experimental results show that FindINT has extremely low overhead and high detection accuracy.

30 September 2024 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
YYYuan Yuan
Yuan Yuan
BYBin Yang
Bin Yang
WSWei Su
Wei Su
HLHaoru Li
Haoru Li
CWChang Wang
Chang Wang
TTTarik Taleb
Tarik Taleb
Wireless edge networks can provide edge services to support various time-critical Internet of Things (IoT) applications, like autonomous vehicles, where cache content updates are significant to maintaining information freshness quantified as information of age (AoI). However, frequent content updates result in high energy consumption at the edge nodes. This paper investigates the cache content updates in wireless edge networks, aiming to ensure information freshness and low energy consumption. To this end, we propose a rainbow deep reinforcement learning-based cache content update scheme (RB-DRN). In the RB-DRN scheme, we first establish a Markov Decision Process (MDP) to characterize the process of cache update. By fully taking advantage of R-Learning empowered Rainbow DQN, we then make optimal strategy to obtain the minimum long-term average overhead associated with energy consumption and information freshness. Extensive simulation results are presented to validate our proposed RB-DRN scheme and also to illustrate that our RB-DRN scheme outperforms the benchmark scheme in terms of information freshness and energy consumption.