Technical Sessions

Session S-1

Edge Computing

10:30 AM — 12:30 PM EDT
Oct 12 Mon, 10:30 AM — 12:30 PM EDT

Sl-EDGE: Network Slicing at the Edge

Salvatore D’Oro (Institute for the Wireless Internet of Things, Northeastern University), Leonardo Bonati (Institute for the Wireless Internet of Things, Northeastern University), Francesco Restuccia (Institute for the Wireless Internet of Things, Northeastern University), Michele Polese (University of Padova, Italy), Michele Zorzi (University of Padova, Italy), Tommaso Melodia (Institute for the Wireless Internet of Things, Northeastern University)

Network slicing of multi-access edge computing (MEC) resources is expected to be a pivotal technology to the success of 5G networks and beyond. The key challenge that sets MEC slicing apart from traditional resource allocation problems is that edge nodes depend on tightly-intertwined and strictly-constrained networking, computation and storage resources. Therefore, instantiating MEC slices without incurring in resource over-provisioning is hardly addressable with existing slicing algorithms. The main innovation of this paper is Sl-EDGE, a unified MEC slicing framework that allows network operators to instantiate heterogeneous slice services (e.g., video streaming, caching, 5G network access) on edge devices. We first describe the architecture and operations of Sl-EDGE, and then show that the problem of optimally instantiating joint network-MEC slices is NP-hard. Thus, we propose near-optimal algorithms that leverage key similarities among edge nodes and resource virtualization to instantiate heterogeneous slices 7.5x faster and within 0.25 of the optimum. We first assess the performance of our algorithms through extensive numerical analysis, and show that Sl-EDGE instantiates slices 6x more efficiently then state-of-the-art MEC slicing algorithms. Furthermore, experimental results on a 24-radio testbed with 9 smartphones demonstrate that Sl-EDGE provides at once highly-efficient slicing of joint LTE connectivity, video streaming over WiFi, and ffmpeg video transcoding.

MVStylizer: An Efficient Edge-Assisted Video Photorealistic Style Transfer System for Mobile Phones

Ang Li (Duke University), Chunpeng Wu (Duke University), Yiran Chen (Duke University), Bin Ni (Quantil Inc.)

Recent research has made great progress in realizing neural style transfer of images, which denotes transforming an image to a desired style. Many users start to use their mobile phones to record their daily life, and then edit and share the captured images and videos with other users. However, directly applying existing style transfer approaches on videos, i.e., transferring the style of a video frame by frame, requires an extremely large amount of computation resources. It is still technically unaffordable to perform style transfer of videos on mobile phones.To address this challenge, we propose MVStylizer, an efficient edge-assisted video style transfer system for mobile phones. To significantly improve the efficiency of style transfer, only extracted key frames from an original video will be directly transformed by a pre-trained deep neural network on edge server_while the rest intermediate frames can be generated by our designed optical-flow-based frame interpolation. In addition, we adopt a federated learning scheme to keep improving the performance of DNN models. Our experiments demonstrated that MVStylizer can generate stylized videos with an even better visual quality compared to the state-of-the-art method while achieving 75.5X speedup for 1920x1080 videos.

Fair Multi-resource Allocation in Mobile Edge Computing with Multiple Access Points

Erfan Meskar, Ben Liang (University of Toronto)

We consider the problem of fair multi-resource allocation for mobile edge computing (MEC) with multiple access points. In MEC, user tasks are uploaded over wireless communication channels to the access points, where they are then processed with multiple types of computing resources. What distinguishes fair multi-resource allocation in the MEC environment from more general cloud computing is that a user may experience different levels of wireless channel quality on different access points, so that the user's channel bandwidth demand is not fixed. Existing resource allocation studies for cloud computing generally consider Pareto Optimality (PO), Envy-Freeness (EF), Sharing Incentive (SI), and Strategy-Proofness (SP) as the most desirable fairness properties. In this work, we show these properties are no longer compatible in MEC, since there exists no resource allocation rule that can satisfy PO+EF+SP or PO+SI+SP. Hence, we propose a resource allocation rule, called Maximum Task Product (MTP), that retains PO, EF, and SI. Extensive simulation driven by Google cluster traces further shows that MTP improves resource utilization while achieving these fairness properties.

Robust Resource Provisioning in Time-Varying Edge Networks

Ruozhou Yu (North Carolina State University), Guoliang Xue, Yinxin Wan (Arizona State University), Jian Tang (Syracuse University), Dejun Yang (Colorado School of Mines), Yusheng Ji (National Institute of Informatics, Japan (NII))

Edge computing is one of the revolutionary technologies that enable high-performance and low-latency modern applications, such as smart cities, connected vehicles, etc. Yet its adoption has been limited by factors including high cost of edge resources, heterogeneous and fluctuating demands, and lack of reliability. In this paper, we study resource provisioning in edge computing, taking into account these different factors. First, based on observations from real demand traces, we propose a time-varying stochastic model to capture the time-dependent and uncertain demand and network dynamics in an edge network. We then apply a novel robustness model that accounts for both expected and worst-case performance of a service. Based on these models, we formulate edge provisioning as a multi-stage stochastic optimization problem. The problem is NP-hard even in the deterministic case. Leveraging the multi-stage structure, we apply nested Benders decomposition to solve the problem. We also describe several efficiency enhancement techniques, including a novel technique for quickly solving the large number of decomposed subproblems. Finally, we present results from real dataset-based simulations, which demonstrate the advantages of the proposed models, algorithm and techniques.

Session Chair

Bin Li (University of Rhode Island)

Enter Zoom
Session S-2

Real-Time Wireless Networking

2:45 PM — 4:15 PM EDT
Oct 12 Mon, 2:45 PM — 4:15 PM EDT

Fresher Content or Smoother Playback? A Brownian-Approximation Framework for Scheduling Real-Time Wireless Video Streams

Ping-Chun Hsieh (National Chiao Tung University), Xi Liu (Texas A&M University), I-Hong Hou (Texas A&M University)

This paper presents a Brownian-approximation framework to optimize the quality of experience (QoE) for real-time video streaming in wireless networks.
In real-time video streaming, one major challenge is to tackle the natural tension between the two most critical QoE metrics: playback latency and video interruption.
To study this trade-off, we first propose an analytical model that precisely captures all aspects of the playback process of a real-time video stream, including playback latency, video interruptions, and packet dropping.
Built on this model, we show that the playback process of a real-time video can be approximated by a two-sided reflected Brownian motion.
Through such Brownian approximation, we are able to study the fundamental limits of the two QoE metrics and characterize a necessary and sufficient condition for a set of QoE performance requirements to be feasible.
We propose a scheduling policy that satisfies any feasible set of QoE performance requirements and then obtain simple rules on the trade-off between playback latency and the video interrupt rates, in both heavy-traffic and under-loaded regimes.
Finally, simulation results verify the accuracy of the proposed approximation and show that the proposed policy outperforms other popular baseline policies.

Optimizing Information Freshness using Low-Power Status Updates via Sleep-Wake Scheduling

Ahmed M. Bedewy (The Ohio State University), Yin Sun (Auburn University), Rahul Singh, Ness Shroff (The Ohio State University)

In this paper, we consider the problem of optimizing the freshness of status updates that are sent from a large number of low-power source nodes to a common access point. The source nodes utilize carrier sensing to reduce collisions and adopt an asychronized sleep-wake strategy to achieve an extended battery lifetime (e.g., 10-15 years). We use \emph{age of information} (AoI) to measure the freshness of status updates, and design the sleep-wake parameters for minimizing the weighted-sum peak AoI of the sources, subject to per-source battery lifetime constraints. When the sensing time is zero, this sleep-wake design problem can be solved by resorting to a two-layer nested convex optimization procedure; however, for positive sensing times, the problem is non-convex. We devise a low-complexity solution to solve this problem and prove that, for practical sensing times that are short and positive, the solution is within a small gap from the optimum AoI performance. Our numerical and NS-3 simulation results show that our solution can indeed elongate the batteries lifetime of information sources, while providing a competitive AoI performance.

Online Control of Random Access with Splitting

Waqas Tariq Toor (Khwaja Fareed University of Engineering and Information Technology), Jun-Bae Seo (Hanyang University), Hu Jin (Hanyang University)

For slotted random access systems, the slotted ALOHA protocol provides the maximum throughput of 0.368 (packets/slot) while in the category of splitting algorithms, the maximum achievable throughput can reach up to 0.487 with the first-come first-serve (FCFS) algorithm. The FCFS algorithm can achieve this maximum throughput only for Poisson traffic arrivals, which limits its application in practical systems. In this paper, we propose a novel online transmission control framework that introduces random splitting for collisions and controls the transmission probabilities optimally at each slot through estimating the number active users in the system. The proposed control is online as it estimates number of active users slot by slot recursively, and thus can adapt to network dynamics. We first reveal that under our proposed framework, the throughput of 0.532 is achievable if the number of users involved in a collision could be known, which serves as a guideline for the upper limit for the random access systems with splitting. Then, for the practical case of unknown collided user number, we show that the proposed algorithm can achieve the maximum throughput of 0.487 for Poisson arrivals while introducing less access delay than FCFS. Thanks to the adaptability to the network dynamics, for non-Poisson arrivals, the proposed algorithm shows much better throughput and delay performance than FCFS.

Session Chair

Xiaowen Gong (Auburn University)

Enter Zoom
Session S-3

Privacy I

4:30 PM — 6:00 PM EDT
Oct 12 Mon, 4:30 PM — 6:00 PM EDT

De-anonymizability of Social Network: Through the Lens of Symmetry

Benjie Miao, Shuaiqi Wang, Luoyi Fu (Shanghai Jiao Tong University), Xiaojun Lin (Purdue University)

Social network de-anonymization, which refers to re-identifying users by mapping their anonymized network to a correlated, unanonymized cross-domain network, is an important problem that has received intensive study in network science. However, it remains less understood how network structural features intrinsically affect whether or not the network can be successfully de-anonymized. To find the answer, this paper offers the first general study on the relation between de-anonymizability and network symmetry. To this end, we propose to capture the symmetry of a graph by the concept of graph homomorphism from abstract algebra. By defining the matching probability matrix, we are able to characterize the de-anonymizability, i.e., the expected number of correctly matched nodes. Specifically, we show that for a graph pair with arbitrary topology, the de-anonymizability is equal to the maximal diagonal sum of the matching probability matrix generated from homomorphisms. Due to the prohibitive cost of enumerating all possible homomorphisms, we further propose an equivalent characterization of de-anonymizability, and accordingly design a sampling algorithm for approximately estimating the de-anonymizability, which significantly reduces the computational cost. Such a general result allows us to theoretically obtain the de-anonymizability of any networks with more specific topology structure. For example, for any classic \ER graph with designated $n$ and $p$, we can represent its de-anonymizability numerically by calculating the local symmetric structure that it contains. Extensive experiments are also performed to validated all our findings. To our best knowledge, this is the first work that rigorously quantifies the relationship between the de-anonymizability and the symmetry property of general networks in a non-asymptotic manner, and thus sheds light on enhancing privacy for real networks design.

Towards Compression-Resistant Privacy-preserving Photo Sharing on Social Networks

Zhibo Wang, Hengchang Guo (Wuhan University), Zhifei Zhang (Adobe Research), Mengkai Song, Siyan Zheng, Qian WangI (Wuhan University), Ben Niu (Institute of Information Engineering, Chinese Academy of Sciences)

The massive photos shared through the social networks nowadays, e.g., Facebook and Instagram, have aided malicious entities to snoop private information, especially by utilizing deep neural networks (DNNs) to learn from those personal photos. To protect photo privacy against DNNs, recent advances adopting adversarial examples could successfully fool DNNs. However, they are sensitive to those image compression methods that are commonly used on social networks to reduce transmission bandwidth or storage space. A recent work proposed to resist JPEG compression, while the compression methods adopted in social networks are black boxes, and variation of compression methods would significantly degrade the resistance.

To the best of our knowledge, this paper gives the first attempt to investigate a generic compression-resistant scheme to protect photo privacy against DNNs in the social network scenario. We propose the Compression-Resistant Adversarial framework ComReAdv that can achieve adversarial examples robust to an unknown compression method. To this end, we design an encoding-decoding based compression approximation model (ComModel) to approximate the unknown compression method by learning the transformation from the original-compressed pairs of images queried through the social network. In addition, we involve the pretrained differentiable ComModel into the optimization process of adversarial example generation and adapt existing attack algorithms to generate compression-resistant adversarial examples. Extensive experimental results on different social networks demonstrate the effectiveness and superior resistance of the proposed ComReAdv to unknown compression as compared to the state-of-the-art methods.

Truthful Mobile Crowd Sensing with Interdependent Valuations

Meng Zhang (Northwestern University) Brian Swenson (Princeton University) Jianwei Huang (The Chinese University of Hong Kong, Shenzhen) H. Vincent Poor (Princeton University)

Mobile crowd sensing (MCS) has been used to enable a wide range of resource-discovery applications by exploiting the ``wisdom'' of many mobile users. However, in many applications, a user's valuation depends on other users' sensory data, which introduces the problem of \textit{interdependent valuations}. This feature can encourage sensory data misreport, hence makes economic mechanisms challenging. While some work has been done to address this problem, the issues of private utility information and communication overheads remain unsolved. In this study, we formulate the first interdependent-valuation model for the resource-discovery MCS systems, aiming to elicit truthful sensory reports and utility information and to maximize expected social welfare. We design a Truthful Sense-And-Bid (T-SAB) Mechanism based on surrogate functions, which can reveal marginal utility information by only requiring each user to submit one-dimensional signaling per resource. We show that the surrogate function and a reward function can limit users' willingness to misreport, when users have small informational sizes, a reasonable condition in large-scale MCS systems. Consequently, our T-SAB Mechanism yields a Perfect Bayesian Equilibrium (PBE) with the efficient allocation outcome, approximate truthfulness, individual rationality, and approximate budget balance. To illustrate the effectiveness of the T-SAB Mechanism, we perform a case study of a cognitive radio network. We demonstrate that the social welfare gain of the T-SAB Mechanism can achieve up to $20\%$ social welfare gain comparing with a benchmark.

Session Chair

Lei Jiao (University of Oregon)

Enter Zoom

Made with in Toronto · Privacy Policy · © 2020 Duetone Corp.