INFOCOM 2019 论文列表

Deep neural networks are becoming increasingly popular in Internet of Things (IoT) applications. Their capabilities of fusing multiple sensor inputs and extracting temporal relationships can enhance intelligence in a wide range of applications. However, one key problem is the missing of adaptation to heterogeneous on-device sensors. These low-end sensors on IoT devices possess different accuracies, granularities, and amounts of information, whose sensing qualities are heterogeneous and vary over time. The existing deep learning frameworks for IoT applications usually treat every sensor input equally over time or increase model capacity in an ad-hoc manner, lacking the ability to identify and exploit the sensor heterogeneities. In this work, we propose SADeepSense, a deep learning framework that can automatically balance the contributions of multiple sensor inputs over time by exploiting their sensing qualities. SADeepSense makes two key contributions. First, SADeepSense employs the self-attention mechanism to learn the correlations among different sensors over time with no additional supervision. The correlations are then applied to infer the sensing qualities and to reassign model concentrations in multiple sensors over time. Second, instead of directly learning the sensing qualities and contributions, SADeepSense generates the residual concentrations that are deviated from the equal contributions, which helps to stabilize the training process. We demonstrate the effectiveness of SADeepSense with two representative IoT sensing tasks: heterogeneous human activity recognition with motion sensors and gesture recognition with the wireless signal. SADeepSense consistently outperforms the state-of-the-art methods by a clear margin. In addition, we show that SADeepSense only imposes little additional resource-consumption burden on embedded devices compared to the corresponding state-of-the-art framework.

Cellular network performance depends heavily on the configuration of its network parameters. Current practice of parameter configuration relies largely on expert experience, which is often suboptimal, time-consuming, and error-prone. Therefore, it is desirable to automate this process to improve the accuracy and efficiency via learning-based approaches. However, such approaches need to address several challenges in real operational networks: the lack of diverse historical data, a limited amount of experiment budget set by network operators, and highly complex and unknown network performance functions. To address those challenges, we propose a collaborative learning approach to leverage data from different cells to boost the learning efficiency and to improve network performance. Specifically, we formulate the problem as a transferable contextual bandit problem, and prove that by transfer learning, one could significantly reduce the regret bound. Based on the theoretical result, we further develop a practical algorithm that decomposes a cell's policy into a common homogeneous policy learned using all cells' data and a cell-specific policy that captures each individual cell's heterogeneous behavior. We evaluate our proposed algorithm via a simulator constructed using real network data and demonstrates faster convergence compared to baselines. More importantly, a live field test is also conducted on a real metropolitan cellular network consisting 1700+ cells to optimize five parameters for two weeks. Our proposed algorithm shows a significant performance improvement of 20%.

Multiuser MIMO (MU-MIMO) empowers access points (APs) with multiple antennas to transmit multiple data streams concurrently to users by exploiting spatial multiplexing. In MU-MIMO, users need to estimate channel state information (CSI) and report it to APs, thus opening a backdoor to attackers who may forge CSI to eavesdrop the content of victims. In this paper, we explore the eavesdropping attack in a novel and practical context in which CSI forgery entangles MU-MIMO user selection in a many-users regime. The attacker hopes to optimize both the eavesdropping opportunity of being selected with the victim and the corresponding decoding quality. We propose new attack and defense mechanisms: (1) USE Attack that enables attackers to achieve near optimal eavesdropping opportunity and high decoding quality through constructing orthogonal CSI against victims followed by stepwise refinements; (2) AngleSec that exploits channel reciprocity for attacker detection without any modification to legacy CSI feedback in which CSI forgery induces a mismatching of downlink and uplink angular spectra at the AP. We implement and evaluate USE Attack and AngleSec in a software defined radio platform WARPv3. Extensive experiments manifest that USE Attack significantly improves the overall eaves-dropping quality compared with state-of-the-art counterparts and AngleSec is able to detect CSI forgery attackers almost for sure.

Wi-Fi backscatter is an emerging technique that enables ultra-low power wireless communications thanks to the simplicity of a backscatter tag. This simplicity drastically reduces communication power. However, this simplicity also removes channel selectivity of the backscatter tag. In this regard, we introduce two issues, violation of the wireless regulations and the waste of resources, in Wi-Fi backscatter networks. To solve these problems, we introduce channel independent packet detection and error vector demodulation. We first design a backscatter receiver accepting Wi-Fi frames on the listening channel as well as adjacent channels, because the backscatter tag responds to all incoming signals regardless of their frequencies. We then investigate how the error vectors of each subcarrier are changed in Wi-Fi backscatter systems. Based on the analysis, we propose a method that translates the error vectors into a backscatter frame. We implement and evaluate our design with commodity 802.11n access points as carrier sources, a software-defined radio as a receiver, and a 2.4 GHz backscatter tag. The results show that channel independent Wi-Fi backscatter is always better than channel dependent approaches.

Recent advances in deep neural networks (DNNs) have substantially improved the accuracy and speed of a variety of intelligent applications. Nevertheless, one obstacle is that DNN inference imposes heavy computation burden to end devices, but offloading inference tasks to the cloud causes transmission of a large volume of data. Motivated by the fact that the data size of some intermediate DNN layers is significantly smaller than that of raw input data, we design the DNN surgery, which allows partitioned DNN processed at both the edge and cloud while limiting the data transmission. The challenge is twofold: (1) Network dynamics substantially influence the performance of DNN partition, and (2) State-of-the-art DNNs are characterized by a directed acyclic graph (DAG) rather than a chain so that partition is greatly complicated. In order to solve the issues, we design a Dynamic Adaptive DNN Surgery (DADS) scheme, which optimally partitions the DNN under different network condition. Under the lightly loaded condition, DNN Surgery Light (DSL) is developed, which minimizes the overall delay to process one frame. The minimization problem is equivalent to a min-cut problem so that a globally optimal solution is derived. In the heavily loaded condition, DNN Surgery Heavy (DSH) is developed, with the objective to maximize throughput. However, the problem is NP-hard so that DSH resorts an approximation method to achieve an approximation ratio of 3. Real-world prototype based on self-driving car video dataset is implemented, showing that compared with executing entire the DNN on the edge and cloud, DADS can improve latency up to 6.45 and 8.08 times respectively, and improve throughput up to 8.31 and 14.01 times respectively.

Assessing the trust between users in a trust social network (TSN) isa critical issue in many applications, e.g., film recommendation,spam detection, and online lending. Despite of various trust assessment methods, a challenge remaining to existing solutions is how to accurately determine the factors that affect trust propagation and trust fusion within a TSN. To address this challenge, we propose the NeuralWalk algorithm to cope with trust factor estimation and trust relation prediction problems simultaneously. NeuralWalk employs a neural network, named WalkNet, to model single-hop trust propagation and fusion in a TSN. By treating original trust relations in a TSN as labeled samples, WalkNet is able to learn the parameters that will be used for trust computation/assessment. Unlike traditional solutions, WalkNet is able to accurately predict unknown trust relations in an inductive manner. Based on WalkNet, NeuralWalk iteratively assesses the unknown multi-hop trust relations among users via the obtained single-hop trust computation rules. Experiments on two real-world TSN datasets indicate that NeuralWalk significantly outperforms the state-of-the-art solutions.

Jobs in modern computing clusters have highly diverse processing durations and heterogeneous resource requirements. In this paper, we consider the problem of online job scheduling for a computing cluster comprised of multiple servers with heterogeneous computation resources, while taking the diversity of resource demands for different jobs into account. Our focus is to achieve a low overall job response time for the system (which is also referred to as the job flowtime) while providing fairness between small and large jobs. Since the job flowtime minimization problem under multiple (even homogeneous) servers are known to be NP-hard, we propose an approximation algorithm to tackle the original online scheduling problem by adopting the notion of fractional job flowtime as a surrogate objective for minimization. We apply Online Convex optimization (OCO) techniques to design the corresponding online scheduling algorithm. More importantly, we show that the dynamic fit of the online version of our approximate algorithm grows only sublinearly with respect to time and derive a bound for its dynamic regret when comparing to its offline counterpart. While the baseline version of our proposed scheduling algorithm assumes the possibilities of job preemption and job migration across different servers, we show that the extent of job preemption and migration can be well controlled by augmenting the objective function of our online convex optimization formulation with the corresponding switching costs.

Screen-camera communication techniques achieve one-way communication with widely-used screen and camera devices. Popular barcode methods use visible spatial patterns to represent data, which has been considered obtrusive to human observers. Recent works borrow ideas from visual light communication (VLC), and use inter-frame pixel change to modulate data. To recover pixel change, the receiver has to record and process video frames. Such video-based extraction has high hardware requirements and lacks reliability. Inspired by VLC-based methods, we propose UnseenCode, an invisible on-screen barcode scheme. It leverages inter-frame embedment from VLC-based methods to embed invisible barcodes into arbitrary on-screen contents. Unlike VLC-based methods, UnseenCode does not require video-based extraction. We propose an image-based extraction method based on cross-component correlation of color images. Any off-the-shelf smartphones with camera capability can be used to read UnseenCode by capturing on-screen contents. We propose the full implementation of UnseenCode for evaluation. Experimental results show that UnseenCode decoding algorithm is reliable and applicable under various screen and camera settings. UnseenCode provides up to 2.5 kbit capacity with less than 5% error rate.

Network performance tomography infers performance metrics on internal network links with end-to-end measurements. Existing results in this domain are mainly Boolean-based, i.e., they check whether or not a link is identifiable, and return the exact value on identifiable links. If a link is not identifiable, Boolean-based solution gives no performance result for the link. In this paper, we extend Boolean-based network tomography to bound-based network tomography where the lower and upper bounds are derived for unidentifiable links. We develop an efficient algorithm to obtain the tightest total error bound, and present a solution that can significantly reduce the total number of measurement paths required for deriving the tightest total error bound. Furthermore, we propose a method to deploy a new monitor over existing ones such that the total error bound could be maximally reduced. Compared to the random monitor deployment and the monitor deployment that maximizes the total number of identifiable links, our monitor deployment method can lead to up to 15 and 2.4 times more reduction on total error bound, respectively.

Voice biometrics is widely adopted for identity authentication in mobile devices. However, voice authentication is vulnerable to spoofing attacks, where an adversary may deceive the voice authentication system with pre-recorded or synthesized samples from the legitimate user or by impersonating the speaking style of the targeted user. In this paper, we design and implement VoicePop, a robust software-only anti-spoofing system on smartphones. VoicePop leverages the pop noise, which is produced by the user breathing while speaking close to the microphone. The pop noise is delicate and subject to user diversity, making it hard to record by replay attacks beyond a certain distance and to imitate precisely by impersonators. We design a novel pop noise detection scheme to pinpoint pop noises at the phonemic level, based on which we establish individually unique relationship between phonemes and pop noises to identify legitimate users and defend against spoofing attacks. Our experimental results with 18 participants and three types of smartphones show that VoicePop achieves over 93.5% detection accuracy at around 5.4% equal error rate. VoicePop requires no additional hardware but only the built-in microphones in virtually all smartphones, which can be readily integrated in existing voice authentication systems for mobile devices.

Information source sampling and update scheduling have been treated separately in the context of real-time status update for age of information optimization. In this paper, a unified sampling and scheduling (S2) approach is proposed, focusing on decentralized updates in multiaccess wireless networks. To gain some insights, we first analyze an example consisting of two-state Markov sources, showing that when both optimized, the unified approach outperforms the separate approach significantly in terms of status tracking error by capturing the key status variation. We then generalize to source nodes with random-walk state transitions whose scaling limit is Wiener processes, the closed-form Whittle's index with arbitrary status tracking error functions is obtained and indexability established. Furthermore, a mean-field approach is applied to solve for the decentralized status update design explicitly. In addition to simulation results which validate the optimality of the proposed S2scheme and its advantage over the separate approach, a use case of dynamic channel state information (CSI) update is investigated, with CSI generated by a ray-tracing electromagnetic software.

We describe a novel protocol for computing the egocentric betweenness centrality of a node when relevant edge information is spread between two mutually distrusting parties such as two telecommunications providers. While each node belongs to one network or the other, its ego network might include edges unknown to its network provider. We develop a protocol of differentially-private mechanisms to hide each network's internal edge structure from the other; and contribute a new two-stage stratified sampler for exponential improvement to time and space efficiency. Empirical results on several open graph data sets demonstrate practical relative error rates while delivering strong privacy guarantees, such as 16% error on a Facebook data set.

FlexRadio, a recent breakthrough in wireless Multi-RF technology, has introduced a new way to unify MIMO and full-duplex into a single framework with a fully flexible design. FlexRadio allows a wireless node to use an arbitrary number of RF chains to support transmission and reception, which makes MIMO and full-duplex subset configurations of FlexRadio. This new architecture has greatly changed the feasibility constraint in wireless networks, which makes the design of high performance MAC layer algorithms even more challenging. First, the RF chain becomes a new resource that needs to be allocated across the network, and the optimal configuration depends not only on the network topology and flow demand, but also on the number of available RF chains at each node. Second, it is not clear how to jointly allocate links and RF chain resources based on the arrival rates and queueing dynamics. In this paper, we introduce a new virtual link model to characterize the feasibility constraint from the perspective of contending RF chain usage. Based on this novel model, a distributed CSMA-like framework is developed to fully leverage the flexibility of RF chain resource allocation.

Conventional vibration sensing systems, equipped with specific sensors (e.g., accelerometer) and communication modules, are either expensive or cumbersome in deployment. In recent years, the community revisits this classic topic by taking advantage of off-the-shelf RFIDs. However, limited by lower reading rate and larger wavelength, current RFID based solutions can only sense low-frequency (e.g. below 100Hz) mechanical vibrations with larger amplitude (e.g. (>) 5mm). To address this issue, this work presents TagSound, an RFID-based vibration sensing system that explores a tag's harmonic backscattering to recover high-frequency and tiny mechanical vibrations accurately. The key innovations are in two aspects: harmonics based sensing and a new recovery scheme. We implement TagSound with USRP platforms. Our comprehensive evaluation shows TagSound can achieve a mean error of 0.37 Hz when detecting vibrations at frequencies below 100Hz, and a mean error of 4.2 Hz even when the vibration frequency is up to 2500Hz.

Mobile edge computing allows wireless users to exploit the power of cloud computing without the large communication delay. To serve data-intensive applications (e.g., augmented reality, video analytics) from the edge, we need, in addition to CPU cycles and memory for computation, storage resource for storing server data and network bandwidth for receiving user-provided data. Moreover, the data placement needs to be adapted over time to serve time-varying demands, while considering system stability and operation cost. We address this problem by proposing a two-time-scale framework that jointly optimizes service (data & code) placement and request scheduling, under storage, communication, computation, and budget constraints. We fully characterize the complexity of our problem by analyzing the hardness of various cases. By casting our problem as a set function optimization, we develop a polynomial-time algorithm that achieves a constant-factor approximation under certain conditions. Extensive synthetic and trace-driven simulations show that the proposed algorithm achieves 90% of the optimal performance.

Running computer vision algorithms on images or videos collected by mobile devices represent a new class of latency-sensitive applications that expect to benefit from edge cloud computing. These applications often demand real-time responses (e.g., <;100 ms), which can not be satisfied by traditional cloud computing. However, the edge cloud architecture is inherently distributed and heterogeneous, requiring new approaches to resource allocation and orchestration. This paper presents the design and evaluation of a latency-aware edge computing platform, aiming to minimize the end-to-end latency for edge applications. The proposed platform is built on Apache Storm, and consists of multiple edge servers with heterogeneous computation (including both GPUs and CPUs) and networking resources. Central to our platform is an orchestration framework that breaks down an edge application into Storm tasks as defined by a directed acyclic graph (DAG) and then maps these tasks onto heterogeneous edge servers for efficient execution. An experimental proof-of-concept testbed is used to demonstrate that the proposed platform can indeed achieve low end-to-end latency: considering a real-time 3D scene reconstruction application, it is shown that the testbed can support up to 30 concurrent streams with an average perframe latency of 32ms, and can achieve 40% latency reduction relative to the baseline Storm scheduling approach.

We consider packet forwarding on a tree with all packets destined for the root, assuming each link may forward at most c ≥ 1 packets each time step. We use the Adversarial Queuing Theory injection model, where a (ρ, σ)-adversary may inject at most σ + ρ · t packets into the network at arbitrary locations during any time interval of length t. The goal is to find a forwarding protocol that minimizes the maximal buffer space required to avoid overflows against a (ρ, σ)-adversary with ρ ≤ c. We consider protocols from the locality viewpoint. A protocol is called d-local if the actions of a node depend only on the current state of nodes at distance at most d. A D-local protocol, where D is the network diameter, is called centralized. It is known that buffers of size Θ(σ + ρ) are necessary and sufficient for centralized protocols. The buffer requirement of O(1)-local protocols was recently proved to be Θ(ρ log D+σ). In this paper, for any d ≥ 2, we describe a d-local algorithm whose buffer space requirement is O (⌈log D/d⌉ ρ + σ). This result is tight, up d to constant factors. In particular, it implies that O(log D) locality is sufficient to achieve the best worst-case performance possible even for centralized algorithms. We also give evidence suggesting that the buffer requirement of a local algorithm designed for trees is good also when the routes do not constitute a single-destination tree.

Liquid leakage detection is a crucial issue in modern industry, which concerns industrial safety. Traditional solutions, which generally rely on specialized sensors, suffer from intrusive deployment, high cost, and high power consumption. Such problems prohibit applying those solutions for large-scale and continuously industrial monitoring. In this work, we present a RFID-based solution, TwinLeak, to detect liquid leakage using COTS RFID devices. Detecting the leakage accurately with coarse-grained RSSI and phase readings of tags has been a daunting task, which is especially challenging when low detection delay is required. Our system achieves these goals based on the fact that the inductive coupling between two adjacent tags is highly sensitive to the liquid leaked between them. Therefore, instead of judging according to the signals of each individual tag, TwinLeak utilizes the relationship between the signals of two tags as an effective feature for leakage detection. Specifically, Twin-Leak extracts discriminative signal features from short segments of signals and instantly identifies leakage using a light-weight classifier. A model-guided method for leakage progress tracking is further devised to simultaneously estimate the leakage volume and rate. We implement TwinLeak, evaluate its performance across various scenarios, and deploy it in a real-world industrial IoT system. In average, TwinLeak achieves a TPR higher than 97.2%, a FPR lower than 0.5%, and a relative property estimation error around 10%, while triggering early alarms after only about 4.6mL liquid leaks.

Database driven dynamic spectrum sharing is one of the most promising dynamic spectrum access (DSA) solution to address the spectrum scarcity issue. In such a database-driven DSA system, the centralized spectrum management infrastructure, called spectrum access system (SAS), collects sensitive operational data of both incumbent users (IUs) and secondary users (SUs), which makes privacy protection critical in this paradigm. However, the few existing solutions rely on online trusted third party, which requires extra infrastructure and brings the risk of single point failure. To address the shortcomings of existing solutions, we propose a privacy enhanced and database-driven dynamic spectrum sharing (PeDSS) framework in this paper, which preserves the privacy for both IUs and SUs in database-driven DSA systems without the need for online trusted third party. Privacy for both IUs and SUs are formally defined and analyzed, and experiment results show that SAS under PeDSS is able to handle a single spectrum request in 0.51 ms on average, which is three orders of magnitude faster than prior arts.