Existing person re-identification (ReID) methods mainly rely on images and videos to match persons across cameras, yet visual data captured by cameras are vulnerable to environmental interferences (e.g. illumination and occlusion) or personal appearance changes, leading to performance degradation under such scenes. Meanwhile, the popularization of Wi-Fi networks has allowed probe requests to be captured for mobile sensing applications such as crowd counting and trajectory estimation. However, the MAC address randomization technique adopted by modern devices breaks the association of probe requests and adversely affects the functionality of these applications. In this paper, we propose MaRPA, the first multimodal training approach that incorporates both videos and Wi-Fi probe requests to simultaneously promote tasks of probe requests association and person ReID. MaRPA first distinguishes among pairwise probe request frames through a contrastive learning model. It then matches video and probe request sequences by exploring their similarities from the position and the vision aspects. Matched videos and probe requests provide complementary information and generate more robust features for both tasks. To evaluate MaRPA, we contribute a new dataset containing synchronous videos and probe requests data for probe requests association and person ReID. Experimental results demonstrate the effectiveness of our approach. For probe requests association, it achieves > 85% discrimination accuracy and > 0.90 V-measure score; for person ReID, it achieves 75.8% mAP and 90.6% Rank-1, improving state-of-the-art video-based ReID methods by over 40%

Recently, deep learning models have been widely explored in recommender systems. Though having achieved remarkable success, the design of task-aware recommendation models usually requires manual feature engineering and architecture engineering from domain experts. To relieve those efforts, we explore the potential of neural architecture search (NAS) and introduce AMEIR for Automatic behavior Modeling, interaction Exploration and multi-layer perceptron (MLP) Investigation in the Recommender system. Specifically, AMEIR divides the complete recommendation models into three stages of behavior modeling, interaction exploration, MLP aggregation, and introduces a novel search space containing three tailored subspaces that cover most of the existing methods and thus allow for searching better models. To find the ideal architecture efficiently and effectively, AMEIR realizes the one-shot random search in recommendation progressively on the three stages and assembles the search results as the final outcome. The experiment over various scenarios reveals that AMEIR outperforms competitive baselines of elaborate manual design and leading algorithmic complex NAS methods with lower model complexity and comparable time cost, indicating efficacy, efficiency, and robustness of the proposed method.

Benefited from rapid developments of deep learning, 3D shape recognition has become a remarkable subject in computer vision systems.The existing methods of multi-perspective views have shown competitive performance in 3D shape recognition.However, they have not yet fully exploited the information among all views of projection.In this paper, we propose a novel Multi-view Moments Embedding Network(MMEN) for capturing multiple moments information.MMEN obtains the similarity between different views and retains the description of the original view by generating moments matrix for representing the general features of the 3D shape.Additionally, we apply the matrix square-root layer to perform a non-linear scaling to the eigenvalues of the moment embedding matrix.We compare the performance of our proposed network with several state-of-the-art models on the ModelNet datasets, and the results of the average instance/class accuracy demonstrate the promising performance of MMEN on 3D shape recognition.

In the era of big data, information on relationships changes along with time, and the graphs of relationships captured at consecutive timestamps form the multiple temporal relational (MTR) network. To identify the relations in the network while preserving the network structure, a common solution is to learn the network representations through network embedding methods, and then build the relations upon the similarity among these representations. However, the existing network embedding methods either focus on a single relation or ignore the correlation between the heterogeneous and homogeneous relations, and thus it is difficult to investigate the multiple temporal features in the network. In this paper, we propose a novel network embedding method, named Homo- Hetero Network Embedding (HHNE), for the MTR networks. The HHNE utilizes the Graph Convolutional Network (GCN) to extract homogeneous features from each temporal relational network and then generates the homo-hetero network embeddings by fusing the single temporal relational features through a Multi- Layer Perceptron (MLP). Therefore, HHNE could capture the multi- dimensional characteristics in the network, including both intra-relation information and inter-relation information. To show the efficiency of HHNE, we conduct experiments in a real-world dataset on predicting new relations in the MTR networks. The result reveals that our method could outperform several legacy network embedding methods and state- of-the-art multi-relational network embedding methods in the task of relation prediction, demonstrating that the proposed embedding method is more suitable for the MTR network.

Computer-aided detection and diagnosis (CAD) have been applied to many departments of medical institutions, and early detection of diseases can prevent serious health loss. Pulmonary diseases generate negative effects on human health, even leading to death. The chest X-ray is a common examination for diagnosis of pulmonary diseases. The experienced radiologist can quickly infer patients' symptoms by screening the chest X-ray image. While in some developing countries or remote rural areas, due to the lack of experienced radiologists or doctors, patients may be misdiagnosed. Many efforts have been spent on developing an effective auxiliary detection system to provide medical workers with evidence on diseases. In particular, detecting pulmonary complication via chest X-ray images is one of the most challenging tasks. In this paper, we transform the pulmonary complications detection task into a multi-binary classification task for each pulmonary pathology, and propose a new classification model, DenXFPN (for X-ray). DenXFPN combines multiple feature maps at different scales extracted through a densely convolutional neural network. Our model achieves 0.827 on the area under the receiver operating characteristic curve (AUC) metric on average, which outperforms the state-of-the-art results on most of all pathologies in the Chest X-ray14 dataset.

People connect to each other over conventional online social networks (OSNs) based on many parameters (common interests, experiences, locations), while the anonymous social networks (ASNs) recommend candidate friends to a user mainly by the location proximity at a coarse-granularity. In this paper, we formulate a fine-grained trajectory-matching prediction problem for friend recommendation in ASNs-what is the likelihood for two users to encounter with each other in the future based on their historical trajectory data? We define the serendipity of two trajectories in both spatial and temporal domains to quantify the similarity between two users' trajectories, and propose an algorithm that recommends candidate friends to a user by determining the similarity of their trajectories. Experiments show that our proposed algorithm can predict the encounter of users quite accurately and it outperforms the conventional algorithms in terms of the precision and consumed time.

Many web applications provide secondary authentication methods, i.e., secret questions (or password recovery questions), to reset the account password when a user's login fails. However, the answers to many such secret questions can be easily guessed by an acquaintance or exposed to a stranger that has access to public online tools (e.g., online social networks); moreover, a user may forget her/his answers long after creating the secret questions. Today's prevalence of smartphones has granted us new opportunities to observe and understand how the personal data collected by smartphone sensors and apps can help create personalized secret questions without violating the users’ privacy concerns. In this paper, we present aSecret-Question based Authenticationsystem, called “Secret-QA”, that creates a set of secret questions on basic of people's smartphone usage. We develop a prototype on Android smartphones, and evaluate the security of the secret questions by asking the acquaintance/stranger who participates in our user study to guess the answers with and without the help of online tools; meanwhile, we observe the questions’ reliability by asking participants to answer their own questions. Our experimental results reveal that the secret questions related to motion sensors, calendar, app installment, and part of legacy app usage history (e.g., phone calls) have the best memorability for users as well as the highest robustness to attacks.

In urban areas, WiFi is the most widely-deployed portal for users to acquire the broadband access. Meanwhile, phishing AP (access point)-a rogue AP that falsifies the SSID (or even the BSSID) of a legitimate corporate AP-has caused many security problems in commodity WiFi networks. Existing research on the phishing AP detection can be divided into two categories: (1) the hardware-based approach usually deploys sensors (sniffers and/or USB-based wireless adapters) and conducts radio frequency (RF) sensing at a large scale to detect the anomaly at link and physical layers; and (2) the measurement-based approach enables a laptop to determine the legitimacy of a given AP by monitoring the RTT (round trip time) of data and/or control messages. However, these approaches require the additional cost on either the hardware deployment, or periodic statistical measurements. In this paper, we present Trident, a context-based reverse authentication method for detecting phishing AP in commodity WiFi networks, which requires no extra hardware deployment or periodic statistical measurements. Specifically, Trident employs a challenge-response protocol that allows a user to (reversely) authenticate an AP by two steps: (1) sending the AP a few questions regarding three user-context features (time, location, traffic) during the user-AP interaction procedure, and (2) examining the answers returned by the AP to determine its legitimacy. Our experimental results reveal that Trident achieves a high reliability rate of 95% and a detection rate of 98% when users are connecting rogue APs in the commodity WiFi network on campus.

The smart phones and their applications add the richness of the VANET applications and increase the requirements on data dissemination support of the networks. This paper focuses on the link reliability model for data dissemination in urban VANETs. First, we extend the analytic model on 1-dimension road to the roads with orthogonal crossings. Then, we verify the link reliability model for the irregularly crossed roads using Monte Carlo simulation. Based on these analytical results, we extend the evolving graph model with reliability (EGwR) on 1-D highway to 2-D urban environment. A data dissemination scheme based on EGwR is designed to verify the model efficacy. The scheme takes into account the macroscopic mobility pattern of vehicles and the impact on the link reliability of individual vehicle movement. Through extensive simulation study, the delivery ratio, latency, and the overhead factor of relay ratio are examined. The results demonstrate that our extended EGwR can effectively improve the data dissemination performance in metropolitan VANETs.

Most applications in urban Vehicular Ad hoc NETworks (VANETs) rely on information sharing, such as real-time traffic information queries, advertisements, etc. However, existing data dissemination techniques cannot guarantee satisfactory performance when a lot of information requests come from all around the network. Because these pieces of information are useful for multiple users located in various positions, it is beneficial to spread the cached copies around. Existing work proposed caching mechanisms and conducted simulations for validation, but there is a lack of theoretical analysis on the explicit caching effects. In this paper, we present the cache coverage ratio as the metric to quantify the caching effects, and theoretical analysis is given based on reasonable assumptions for urban VANETs, through which we find the affecting factors include vehicle density, transmission range, ratio of caching vehicles, etc. We deduce the quantitative relationship among them, which have similar forms as the cumulative density function of an exponential distribution. We conduct intensive simulations, which verify the theoretical analysis results match quite well with the simulated reality under different scenarios.

Abstract Most applications in urban vehicular ad hoc networks (VANETs) rely on information sharing, such as real‐time traffic information queries, and advertisements. However, existing data dissemination techniques cannot guarantee satisfactory performance when amounts of information requests come from all around the network. Because these pieces of information are useful for multiple users located in various positions, it is beneficial to spread the cached copies around. Existing work proposed caching mechanisms and conducted simulations for validation, but there is a lack of theoretical analysis on the explicit caching effects. Because of the complex urban environment and high mobility of vehicles, quantifying the caching effects on the VANET performance is quite challenging. We present the cache coverage ratio as the metric to measure the caching effects, and theoretical analysis is given based on reasonable assumptions for urban VANETs, through which we find the affecting factors include vehicle density, transmission range, and ratio of caching vehicles. We deduce the quantitative relationship among them, which have similar forms as the cumulative density function of an exponential distribution. We also consider the impact of vehicle mobility to predict the future cache effect on surrounding roads of the caching area. We conduct intensive simulations, which verify that the theoretical analysis results match quite well with the simulated reality under different scenarios. Copyright © 2015 John Wiley & Sons, Ltd.

The Radio Resource Control (RRC) state machine has been widely studied to improve application performance and reduce the energy consumption of smart phones. We observe that even a smart phone remains in the RRC state, there may still exist some significant performance change if it simultaneously carry on voice and data services. For example, the Round Trip Time (RTT) will get larger when a voice call comes upon, and may get even larger after finishing the voice call. We conduct in-depth study and find out the primary cause of those phenomena - smart phones may get different physical channels even in the same RRC state. In this paper, we define new states according to which physical channel is allocated to smart phones, then discover a fine-grained state machine which is hidden in the physical layer of user plane, which can explain those phenomena very well. We construct the hidden fine-grained state machine for two carriers, one only supports 3G network in most area and the other supports 4G network. The state machine can help us analyze the interference of voice to data. Using the fine-grained state machine, we also offer some suggestions on how to avoid those performance loss both from the perspective of carriers and users.

In this paper, we present TapLock as a smartphone password system that exploits the finger tap events on capacitive touch screens for increasing the password's resilience to shoulder-surfing attacks (where the password input by a user can be easily observed by a bystander over the user's shoulder). TapLock captures the size and the axis length of the finger touch area on the phone screen for creating a password, which cannot be easily observed by a shoulder surfer. Our user study shows that TapLock has several advantages over existing smartphone password systems, including its strong attack resilience, small authentication delay, and haptic input feedback that improves the usability.

Named Data Networking (NDN) is a data-centric architecture designed for the future Internet. Existing works show that NDN brings significant performance improvement for typical content-centric applications, and can also fit the mobile environment well. However, directly applying NDN to Vehicular Ad hoc NETworks (VANETs) is confronted with great challenges due to the high mobility of vehicles. Most applications in VANETs are relied on data dissemination mechanisms. Therefore, we aim to improve the performance of NDN packet forwarding for the efficient content delivery in urban VANET scenarios. Specifically, we introduce the geo-location information to the NDN forwarding plane, and propose a geo-based forwarding strategy to make NDN fit the urban VANETs. Simulation results show our strategy can achieve 27% ~ 75% higher request success ratio, and 40% ~ 80% lower delay compared with the default NDN strategy in urban scenarios with different vehicle densities.

Abstract Multi‐interface multi‐channel can be used to reduce the channel interference and improve the network capacity for multi‐hop wireless ad‐hoc networks, but using multi‐channel will impact the network connectivity, which is an important quality of service factor in VANET. The basic multi‐channel assignment scheme in our paper keeps the network connectivity resilient to the mobility. On the basis of this method, we design a dynamic channel assignment algorithm that can dynamically switch the channels to the less busy ones by monitoring the channel usages. In order to survive the highly mobile scenarios in VANET, the algorithm is designed in a fully distributed way with low overhead. Simulations show that our method can notably improve the throughput of VANETs. And the performance keeps steady with various network configurations. Copyright © 2014 John Wiley & Sons, Ltd.

As streaming media services in wireless networks become popular, it is necessary to study new multicast algorithms with low latency to meet real-time requirements. Wireless mesh networks (WMNs) offer potential solutions to problems of streaming media services. This paper proposes a new two-tier architecture assisted by gateways for WMNs. Based on this architecture, multicast tree establishment algorithm with bandwidth prediction model is introduced, in which links with better bandwidth conditions will be chosen as tree branches to shorten latency to the largest extent. Furthermore, routing algorithm called Gateway Assisted Multicast Algorithm towards Low Latency for Wireless Mesh Networks (GAMLL) is designed to support real-time streaming media services, in which routing maintenance and recovery is also discussed. Simulation results show GAMLL with prediction model works better in VoIP and video transmission services than On-Demand Multicast Routing Protocol (ODMRP), which has relative low latency in wireless networks. Compared to ODMRP, GAMLL reduces average latency and maximum latency by up to 28% and 52% respectively.

Data dissemination is an important technology for most VANET applications. Due to the physical factors like traffic lights, the density of vehicles is usually uneven. Areas near intersections often have a high node density while areas near the center of streets can have a very low node density. The uneven distribution of nodes and the movement pattern of vehicles cause intermittent connections and redundant transmissions, which we call MIRT (Mobility Induced Redundant Transmission). Existing data dissemination algorithms did not pay enough attention to MIRT and the intermittent connections of the network, which makes them unsuitable for realistic VANET scenarios. In this paper, we propose the Bloom Filter based Buffering Data Dissemination (BFBD) algorithm. BFBD copes with MIRT by using decayed bloom filter as the structure to record the data dissemination history and handles the intermittent connectivity by a carefully designed buffering mechanism. Extensive simulations show the performance of BFBD is superior over contrastive algorithms in terms of reliability, dissemination efficiency and delay.

We present a link capacity estimation model and a hybrid routing scheme for wireless mesh networks which aims at improving throughput. The multi-source interference between links and frame capture behaviors of nodes are analyzed. Our model provides estimated values which can indicate the relative quality of wireless links. And we present a routing metric which combines multiple factors based on the estimated values. The proposed routing scheme which is independent of special routing protocols adopts different strategies for the traffic of accessing the Internet and the end-to-end traffic. Besides, we introduce a new gateway-assisted routing technique to further improve the performance. Simulation results show that the estimation model provides accurate estimated values enough for route selection, and the routing scheme improves end-to-end throughput by 185% compared with legacy AODV[14] protocol.

Nowadays, Vehicular Ad hoc NETworks (VANETs) attract more and more attentions both from academia and industry. Although it has achieved much success in the research field, large-scale deployments of VANETs are still lacking. One barrier facing us is that both governments and companies are doubtful about the performance of VANETs in large-scale deployment in real world. The connectivity of VANETs is a factor closely related to the performance of VANETs. We give a thorough theoretical analysis of the VANET connectivity using bond percolation model and Bollobas model in different scenarios. We discover the quantitative relationship among network connectivity, vehicle density and transmission range. Given the vehicle density, we can calculate the minimum transmission range to achieve good network connectivity. Simulations conducted in a large scenario validate our analysis. Our results not only give us insights about the properties of the network topology, but also have great meanings in real world, which can guide the deployment of VANETs.

The vehicular ad hoc network (VANET) has attracted a lot of interest recently. However, traditional VANET is just an instantiation of MANET in which nodes are treated equally for data delivery. We first analyze the unique features of urban VANET that vehicles have different types, and move as clusters due to the influence of traffic lights. Then a two tier architecture called Mobile Infrastructure Based VANET (MIVANET) is proposed. In this architecture, the buses constitute a mobile backbone for data delivery while the low tier is composed of ordinary cars and passengers. MI-VANET will not only bring the benefit that ordinary cars do not have to forward packets for other nodes, but also improve the network connectivity. The corresponding Mobile Infrastructure Registering (MIRG) and Mobile Infrastructure Routing (MIRT) algorithms are also presented. The bus line information is made full use in MIRT. Simulation results show that there is a 40-55% improvement in delivery ratio while the throughput is even doubled compared to GPSR in traditional VANET.

Geographical source routing is a promising routing technique for VANETs, due to its adaptability for network dynamics and ability to handle topology holes. In traditional geographical source routing algorithms a best-known neighbor, typically the neighbor closest to the next junction in a greedy fashion, is designated as the next hop. This approach may cause two drawbacks: (1) the designated neighbor might not receive the packet correctly and (2) non-neighbor nodes are never given opportunities to do forwarding. Due to the broadcast nature of wireless media, non-neighbor nodes can overhear the transmission and may offer better forwarding progress than the designated neighbor. The above two factors both degrades the end-to-end delivery ratio. In this paper, we resolve the degradation of delivery ratio by introducing the concept of opportunistic routing to geographical source routing. A new routing protocol, named Geographical Opportunistic Source Routing (GOSR), is developed. GOSR allows non-neighbor nodes as well as the bestknown neighbor to become a forwarder. The notification cost of opportunistic routing is minimized by enforcing a scope from which candidate forwarders are selected. Defer timers are adopted to avoid conflicts caused by simultaneous transmissions by nodes in the designated scope. Extensive simulation studies are performed and the results show that GOSR can substantially reduce hop count and also improve end-to-end delivery ratio remarkably.

In this paper, we design and implement an applied wireless mesh network named MEMO with transparent client access support and mobile node mobility management. Unlike traditional design, our proposed solution is based on MAC layer triggered mechanism and special routing scheme called AODV_MEMO which combines routing and mobility management into a single solution. The advantages of AODV-MEMO solution include: (1) no modification is required on mobile nodes; (2) mobility management is accomplished in distributed way, no central location server is needed; (3) cross- layer interaction between MAC and Routing Layer effectively reduces the handover time of mobile nodes roaming. In addition, an effective gateway function has been implemented based on the AODV-MEMO routing. Real test-bed experiment proves our design is feasible and efficient.

In this paper, we propose a novel approach to emulate an arbitrary wireless multi-hop network scenario in a fixed wireless testbed, in which every node is equipped with a variable attenuator. We formulate the core problem of wireless multi-hop network scenario emulation as a constrained optimization problem and propose an optimum algorithm