INFOCOM 2019 论文列表

Machine Learning (ML) is under intense investigation in optical networks as it promises to lead to automation of a variety of management tasks, as amplifier gain equalization, fault recognition, Quality of Transmission (QoT) estimation, and many others. Though several studies focus on each of these specific tasks, the integration of ML-based estimations inside Routing and Spectrum Assignment (RSA) is still largely unexplored.This paper moves towards such integration. We develop a framework that leverages the probabilistic outputs of a ML-based QoT estimator to define the reach constraints in an Integer Linear Programming (ILP) formulation for RSA in an elastic optical network. In this integrated procedure, the RSA problem is solved iteratively by updating the reach constraints based on the outcome of a QoT estimator, to exclude lightpaths with unacceptable QoT. In our numerical evaluation, the proposed integrated method achieves savings in spectrum occupation up to 30% (around 20% on average) compared to traditional ILP-based RSA approaches with reach constraints based on margined analytical models.

In this paper, we propose a task scheduling strategy, which can achieve image feature extraction on edge while ensuring privacy. Our task scheduling strategy applies to a fairly popular privacy-preserving Scale-Invariant Feature Transform SIFT scheme, where images to be processed are firstly randomly split into two portions for encryption and transmitted to two different edge nodes for feature extraction. Then, in the edge, our task scheduling strategy will re-assign these two portions to proper edge nodes for processing. During the whole process, two portions of the same image should not be assigned to the same edge node in order to preserve privacy. We show that this privacy constraint can be enforced through constructing a pairwise Markov chain, and carefully designing system states and transition probabilities. We further formulate the whole task scheduling problem as a stochastic latency minimization problem and solve it by converting it into a linear programming problem. Simulation results show that our proposed task scheduling strategy can achieve lower latency than baseline strategies while satisfying the privacy constraint.

Accurate modeling of network traffic has a wide variety of applications. In this paper, we propose Network Transmission Point Process (NTPP), a probabilistic deep machinery that models the traffic characteristics of hosts on a network and effectively forecasts the network traffic patterns, such as load spikes. Existing stochastic models relied on the network traffic being self-similar in nature, thus failing to account for traffic anomalies. These anomalies, such as short-term traffic bursts, are very prevalent in certain modern-day traffic conditions, e.g. datacenter traffic, thus refuting the assumption of self-similarity. Our model is robust to such anomalies since it effectively leverages the self-exciting nature of the bursty network traffic using a temporal point process model.On seven diverse datasets collected from the fields of cyberdefense exercises (CDX), website access logs, datacenter traffic, and P2P traffic, NTPP offers a substantial performance boost in predicting network traffic characteristics against several baselines, ranging from forecasting the network traffic volume to detecting traffic spikes. We also demonstrate an application of our model to a caching scenario, showing that it can be used to effectively lower the cache miss rate.

Multi-user MIMO (MU-MIMO) is a technique in 802.11ac and 802.11ax that improves spectral efficiency by allowing concurrent communication between one AP and multiple clients. In practice, the expected gain is not always achieved and is sometimes even negative. Using a commodity 802.11ac AP, we experimentally determine that the inclusion of clients either in motion or with low SNR can cause throughput below that of single-user transmissions. We then propose a pre-screening algorithm using reinforcement learning to predict if a client can benefit from participating in MU-MIMO. Our algorithm is based on a sequence of channel state information (CSI), SNR, and client device type, and can automatically adapt to the motion of individual clients. Experimental results using a commodity AP show that the additional implementation of the pre-screening algorithm alone, without otherwise modifying MU-MIMO client grouping or link parameter selection algorithms, can improve system throughput by up to 40% when half of the clients are moving. Over 20% throughput improvement is maintained when between 25% to 75% of the clients are moving.

Traditionally, Network Function Virtualization uses Service Function Chaining (SFC) to place service functions and chain them with corresponding flows allocation. With the advent of Edge computing and IoT, a retiable composition of latency-sensitive SFCs is needed to support applications in geo-distributed cloud infrastructures. However, the optimal SFC composition in this case becomes the NP-hard integer multi-commodity-chain flow (MCCF) problem that has no known approximation guarantees. In this paper, we present a novel practical and near optimal SFC composition approach for geo-distributed cloud infrastructures that also admits end-to-end network QoS constraints such as latency, packet loss, etc. Specifically, we propose a novel metapath composite variable approach that reaches 99% optimality on average and takes seconds for practically sized integer MCCF problems of US Tier-1 (~300 nodes) and regional (~600 nodes) infrastructure providers' topologies. To ensure reliability, we compose SFCs with capacity chance-constraints and backup policies. Using trace-driven simulations comprising of challenging disaster-incident conditions, we show that our solution composes twice as many SFCs than the state-of-the-art network virtualization methods.

Network consensus optimization has received increasing attention in recent years and has found important applications in many scientific and engineering fields. To solve network consensus optimization problems, one of the most well-known approaches is the distributed gradient descent method (DGD). However, in networks with slow communication rates, DGD's performance is unsatisfactory for solving high-dimensional network consensus problems due to the communication bottleneck. This motivates us to design a communication-efficient DGD-type algorithm based on compressed information exchanges. Our contributions in this paper are three-fold: i) We develop a communication-efficient algorithm called amplified-differential compression DGD (ADC-DGD) and show that it converges under any unbiased compression operator; ii) We rigorously prove the convergence performances of ADC-DGD and show that they match with those of DGD without compression; iii) We reveal an interesting phase transition phenomenon in the convergence speed of ADC-DGD. Collectively, our findings advance the state-of-the-art of network consensus optimization theory.

One critical challenge of managing modern data center networks lies in that existing network protocols provide limited visibility on the internal routing and forwarding decisions made by the control plane, leading to difficulties on fast diagnosis and identification of root causes for performance bugs and anomalies. In this paper, we develop and evaluate a “debugging mode” for packet forwarding, where we demonstrate a possible design space by introducing a programmable header field into data packets used for diagnosis purposes. These headers can be manipulated by routers in intermediate hops to perform tracing and diagnosis operations, thereby providing much greater visibility on the control plane and data plane operations. To make this design scalable and feasible, we exploit the software APIs provided by the latest software-defined networking (SDN) technologies, where the network control plane is separated from the underlying data plane, so that we can reprogram the network forwarding functions dynamically. Compared to existing alternative approaches, our approach is adaptive and programmable, allowing dynamic and on-demand receiver-side decoding with extremely low overhead. We emphasize that as this “debugging mode” can be enabled and disabled by network managers as demanded, it introduces zero overhead to normal traffic if everything is operating as expected. Our evaluation results on a real SDN network testbed demonstrate the effectiveness of the proposed approaches.

This paper demonstrates the feasibility of a side-channel attack to infer keystrokes on touch screen leveraging an off-the-shelf smartphone. Although there exist some studies on keystroke eavesdropping attacks on touch screen, they are mainly direct eavesdropping attacks, i.e., require the device of victims compromised to provide side-channel information for the adversary, which are hardly launched in practical scenarios. In this work, we show the practicability of an indirect eavesdropping attack, KeyListener, which infers keystrokes on QWERTY keyboards of touch screen leveraging audio devices on a smartphone. We investigate the attenuation of acoustic signals, and find that a user's keystroke fingers can be localized through the attenuation of acoustic signals received by the microphones in the smartphone. We then utilize the attenuation of acoustic signals to localize each keystroke, and further analyze errors induced by ambient noises. To improve the accuracy of keystroke localization, KeyListener further tracks finger movements during inputs through phase change and Doppler effect to reduce errors of acoustic signal attenuation-based keystroke localization. In addition, a binary tree-based search approach is employed to infer keystrokes in a context-aware manner. The proposed keystroke eavesdropping attack is robust to various environments without the assistance of additional infrastructures. Extensive experiments demonstrate that the accuracy of keystroke inference in top-5 candidates can approach 90% with a top-5 error rate of around 6%, which is a strong indication of the possible user privacy leakage of inputs on QWERTY keyboard.

The ability to support continuous network configuration updates is an important ability for enabling Software Defined Networks (SDN) to handle frequent or bursty changes. Current solutions for updating SDN configurations focus on one single update at a time, leading to slow, sequential (i.e., blocking) update execution. In this paper, we develop update algebra, a novel, systematic, theoretical framework based on abstract algebra, to enable continuous, non-blocking, fast composition of multiple updates. Specifically, by modeling each data-plane operation in the set of data-plane operations to be executed by an update as a set-theoretical projection, update algebra defines novel operation composition so that the number of projections for the same match remains constant regardless of the number of updates to be composed, leading to substantial performance benefits. Specifying the dependencies of the data-plane operations in updates as a subset of a free monoid in the general case and as partial ordering for basic consistency, update algebra defines update composition that preserves consistency, even under partially-executed updates, to guarantee correctness. We conduct asymptotic analysis, extensive benchmarking using a real controller, and integration with a real application to demonstrate the benefits of update algebra. In particular, our asymptotic analysis demonstrates that in independent-update dominant settings, update completion time of update algebra remains asymptotically constant despite growth of the number of updates to be executed. Our benchmarking shows that update algebra can achieve 16x reduction in update latency even in settings with an update arrival rate of only 1. 6/s. Our integration with Hedera, a real SDN traffic engineering application, shows that update algebra can reduce average link bandwidth utilization by 30% compared with sequential updates.

When a program is executed on an untrusted cloud, the confidentiality of the program's logics needs to be protected. Control flow obfuscation is a direct approach to obtain this goal. However, existing methods in this direction cannot achieve both high confidentiality and low overhead. In this paper, we propose CFHider, a hardware-assisted method to protect the control flow confidentiality. By combining program transformation and Intel Software Guard Extension (SGX) technology, CFHider moves branch statement conditions to an opaque and trusted memory space, i.e., the enclave, thereby offering a guaranteed control flow confidentiality. Based on the design of CFHider, we developed a prototype system targeting on Java applications. Our analysis and experimental results indicate that CFHider is effective in protecting the control flow confidentiality and incurs a much reduced performance overhead than existing software-based solutions (by a factor of 8.8).

Motivated by practical scenarios in areas such as Mobile Edge Computing (MEC) and Content Delivery Networks (CDNs), we study online file caching on multiple caches, where a file request might be relayed to other caches or bypassed directly to the memory when a cache miss happens. We take the relaying, bypassing and fetching costs altogether into consideration. We first show the inherent difficulty of the problem even when the online requests are of uniform cost. We propose an O(log K)-competitive randomized algorithm Camul and an O(K)-competitive deterministic algorithm Camul-Det, where K is the total number of slots in all caches. Both online algorithms achieve asymptotically optimal competitive ratios, and can be implemented efficiently such that each request is processed in amortized constant time. We conduct extensive simulations on production data traces from Google and a benchmark workload from Yahoo. It shows that our algorithms dramatically outperform existing schemes, i.e., reducing the total cost by 85% and 43% respectively compared with important baselines and their strengthened versions with request relaying. More importantly, Camul achieves such a good total cost without sacrificing other performance measures, e.g., the hit ratio, and can perform consistently well on various settings of experiment parameters.

The discovery of Wireless Power Transfer (WPT) technologies makes charging more convenient and reliable. Among all the existing WPT technologies, directional WPT is more efficient and has been successfully applied to supply energy for wireless rechargeable sensor networks (WRSNs). However, the state-of-the-art methods ignore the anisotropic energy receiving property of rechargeable sensors, resulting in energy wastage. In order to address this issue, in this paper, we point out that the received energy of a sensor is not only relative to the distance, but also relative to the angle between the sensor and the charger's orientation in directional WPT. Towards this end, we derive a pragmatic energy transfer model verified by experiments. In particular, we focus on a Minimal chArging Delay (MAD) problem to reduce charging delays. To obtain the optimal solution, we formulate the problem as a linear programming problem. Moreover, we introduce a method of charging power discretization, which significantly reduces the search space and bounds the performance gap to the optimal one with a 1/1-ϵ2approximation ratio. Besides, a merging method is introduced for a more practical application scenario. Finally, we demonstrate that our methods outperform the Set Cover baseline method by an average of 34.2% through simulations and experiments.

This paper addresses a fundamental challenge in cloud computing, that of learning an economical yet robust reservation, i.e. reserve just enough resources to avoid both violations and expensive over provisioning. Prediction tools are often inadequate due to observed high variability in CPU and memory workload. We propose a novel model-free approach that has its root in online learning. Specifically, we allow the workload profile to be engineered by an adversary who aims to harm our decisions, and we investigate a class of policies that aim to minimize regret (minimize losses with respect to a baseline static policy that knows the workload sample path). Then we propose a combination of the Lyapunov optimization theory [1] and a linear prediction of the future based on the recent past, used in learning and online optimization problems, see [2]. This enables us to come up with a no regret policy, i.e., a policy whose cost difference to the benchmark and violation constraint residual both grow sublinearly in time, and hence become amortized over the horizon. Our policy has then “no regret and eventually learns the minimum cost reservation subject to a time-average constraint for violations.

The computing frontier is moving from centralized mega datacenters towards distributed cloudlets at the network edge. We argue that cloudlets are well-suited for participation in Emergency Demand Response (EDR) programs due to their enormous energy consumption and flexible workload distribution, while existing EDR mechanisms for clouds and colocation datacenters are not suitable for cloudlets. We propose a novel online market mechanism, EdgeEDR, to incentivize cloudlets to participate in EDR, featuring multiple cloudlet-specific designs. At a high level, we observe that cloudlet operators can dynamically switch on/off entire cloudlets to compensate for the energy reduction required by the power grid. We formulate a long-term social cost minimization problem and decompose it into a series of one-round procurement auctions. In each auction instance, we propose to let the cloudlet tenants bid with cost functions of their service quality degradation tolerance, and let the cloudlet operator choose the service quality, allocate the workload, and shut down the cloudlets. Via rigorous analysis, we exhibit that our bidding policy is individually rational and truthful; our workload distribution algorithm has near-optimal performance in each auction; and our overall online algorithm achieves a provable competitive ratio. We further confirm the performance of our mechanism through extensive trace-driven simulations.

With the proliferation of smartphones, a novel sensing paradigm called Mobile Crowd Sensing (MCS) has emerged very recently. However, the attacks and faults in MCS cause a serious false data problem. Observing the intrinsic low dimensionality of general monitoring data and the sparsity of false data, false data detection can be performed based on the separation of normal data and anomalies. Although the existing separation algorithm based on Direct Robust Matrix Factorization (DRMF) is proven to be effective, requiring iteratively performing Singular Value Decomposition (SVD) for low-rank matrix approximation would result in a prohibitively high accumulated computation cost when the data matrix is large. In this work, we observe the quick false data location feature from our empirical study of DRMF, based on which we propose an intelligent Light weight Low Rank and False Matrix Separation algorithm (LightLRFMS) that can reuse the previous result of the matrix decomposition to deduce the one for the current iteration step. Our algorithm can largely speed up the whole iteration process. From a theoretical perspective, we validate that LightLRFMS only requires one round of SVD computation and thus has very low computation cost. We have done extensive experiments using a PM 2.5 air condition trace and a road traffic trace. Our results demonstrate that LightLRFMS can achieve very good false data detection performance with the same highest detection accuracy as DRMF but with up to 10 times faster speed thanks to its lower computation cost.

The device-to-device (D2D) communication is an efficient mobile-data offloading technique to significantly improve the cellular base station (BS) spectrum efficiency in the fifth generation (5G) wireless networks by enabling mobile users to directly receive the multimedia data from a nearby mobile users through D2D communication. On the other hand, D2D multimedia-data offloading imposes the new modeling challenges in characterizing the statistical delay-bounded quality of service (QoS) provisioning over two-hop tandem wireless links when a mobile user receives its requested data from the cellular BS and relays it to the next mobile user which is requesting the same data via D2D communications. Thus, one of the most important challenges of the D2D multimedia-data offloading architecture lies in how to satisfy the statistical delay-bounded QoS requirements over the two-hop wireless links which consists of two tandem single-hop links. To tackle these challenges, we propose the overall two-hop wireless link QoS provisioning schemes based on its two individual single-hop statistical QoS requirements. We derive the overall effective capacity's expression of the two-hop tandem wireless links as a function of the single-hop's effective capacities and prove that the overall equivalent effective capacity of the two-hop wireless link is upper bounded by the single-hop effective capacity with the smaller QoS exponent. Moreover, we derive the upper-bound and lower-bound of the equivalent QoS exponent for the overall two-hop wireless links and also show its monotonicity with respect to single-hop QoS exponents. To show the network performance's improvement after integrating the cellular networks and D2D offloading, we compare the scalability of the BS-based cellular wireless networks with and without D2D offloading. We also evaluate and validate our proposed two-hop wireless link statistical delay bounded QoS provisioning schemes over multimedia D2D offloading architecture through numerical analyses.

KPI (Key Performance Indicator) anomaly detection is critical for Internet-based services to ensure the quality and reliability. However, existing algorithms' performance in reality is far from satisfying due to the lack of sufficient KPI anomaly data to help train and evaluate these algorithms. In this paper, we argue that labeling overhead is the main hurdle to obtain such datasets. Thus, we novelly propose a semi-automatic labelling tool called Label-Less, which minimizes the labeling overhead in order to enable an ImageNet-like large-scale KPI anomaly dataset with high-quality ground truth. One novel technique in Label-Less is robust and rapid anomaly similarity search, which saves operators from scanning and checking the long KPIs back and forth for abnormal patterns or label consistency. In our evaluations using 30 real KPIs from a large Internet company, our anomaly similarity search achieves the best F-score of 0.95 on average, and a real-time per-KPI response time (less than 0.5 second). Overall, the feedback from deployment in practice shows that Label-Less can reduce operators' labeling overhead by more than 90%.

The great market success of Blockchain makes it an extremely valuable target for attackers. A well-known attack in Blockchain is the forking attack, where divergent blockchains are produced for inserting some new features to facilitate security breaches. The state-of-the-art works mostly focus on how to detect attacks in real-time transactions, which is in hindsight and cannot deter the forking attack from the root. To take precautions, we employ the large deviation theory to study the vulnerability of blockchain networks incurred by intentional forks from a micro point of view, boosting forward-looking and strategic planning mechanisms for resisting the forking attack. Our study is fine-grained, because it offers not only the vulnerability probability of a blockchain network but also its decay speed, through which we find setting the parameter related to the robust level has more power than enhancing the computer power in speeding up the failure of attacks. This finding is valuable since it renders an opportunity to improve the robustness of a blockchain network in a cost-efficient way. Our analysis is complementary, since it studies both the impacts of the computational power as well as the number of confirmations on the vulnerability of a blockchain network, providing a theoretical basis to design reasonable schemes for invigorating a blockchain network from technical as well as managerial levels. Extensive experiments carried out on a large-scale cloud platform running the Ethereum protocol show the experimental and analytical results match well, verifying the effectiveness of our analysis.

In this paper we investigate the allocation of resources to slices on the radio interface of one cell. In particular, we develop a detailed stochastic model of the behaviour of the sliced cell radio access, including most features of the standard access procedures. Our model allows the computation of the throughput achieved by each slice, as well as the distribution of delays for each slice. The availability of a model capable of accurately predicting the performance achieved by services using different slices as a function of the cell parameters is extremely important for the automated run time management of the cell and for the correct setting of its parameters.Specifically, while our model can cope with a number of slices, we focus on the case of one cell comprising one slice for human type communications and one slice for machine type communications, and we discuss relevant emerging behaviours in the slices performance, as functions of the cell parameters.We validate the analytical predictions by comparison against the estimates of a detailed simulator, proving the accuracy of the model. Our model turns out to be very effective in providing insight and guidelines for allocation and management of resources in cells hosting slices carrying traffic derived from services with different characteristics and performance requirements.