Human activity recognition with smartphone sensors using deep learning neural networks

Human activities are inherently translation invariant and hierarchical. Human activity recognition (HAR), a field that has garnered a lot of attention in recent years due to its high demand in various application domains, makes use of time-series sensor data to infer activities. In this paper, a deep convolutional neural network (convnet) is proposed to perform efficient and effective HAR using smartphone sensors by exploiting the inherent characteristics of activities and 1D time-series signals, at the same time providing a way to automatically and data-adaptively extract robust features from raw data. Experiments show that convnets indeed derive relevant and more complex features with every additional layer, although difference of feature complexity level decreases with every additional layer. A wider time span of temporal local correlation can be exploited (1 × 9–1 × 14) and a low pooling size (1 × 2–1 × 3) is shown to be beneficial. Convnets also achieved an almost perfect classification on moving activities, especially very similar ones which were previously perceived to be very difficult to classify. Lastly, convnets outperform other state-of-the-art data mining techniques in HAR for the benchmark dataset collected from 30 volunteer subjects, achieving an overall performance of 94.79% on the test set with raw sensor data, and 95.75% with additional information of temporal fast Fourier transform of the HAR data set.     

Detecting proximity events in sensor networks

Sensor networks are often used to perform monitoring tasks, such as animal and vehicle tracking, or the surveillance of enemy forces in military applications. In this paper we introduce the concept of proximity queries, which allow us to report interesting events, observed by nodes in the network that lie within a certain distance from each other. An event is triggered when a user-programmable predicate is satisfied on a sensor node. We study the problem of computing proximity queries in sensor networks and propose several alternative techniques that differ in the number of messages exchanged by the nodes and the quality of the returned answers. Our solutions utilize a distributed routing index, maintained by the nodes in the network, that is dynamically updated as new observations are obtained by the nodes. This distributed index allows us to efficiently process multiple proximity queries involving several different event types within a fraction of the cost that a straightforward evaluation requires. We present an extensive experimental study to show the benefits of our techniques under different scenarios using both synthetic and real data sets. Our results demonstrate that our algorithms scale better and require significantly fewer messages compared to a straightforward execution of the queries.     

Node discovery in networks

This paper abstracts the problem of network nodes discovering one another in a network of unknown size using all-to-all gossip. The problem is studied in terms of evolving directed graphs where each vertex represents a participating node and each edge represents one node’s knowledge about another. Ideally, such a graph has diameter one, i.e., each node knows all others. Nodes share their knowledge by sending gossip messages. Gossip among the nodes allows them to discover one another, decreasing the diameter of the graph. Here this problem is considered in several synchronous settings under different assumptions about the ability of the participating nodes to communicate. Specifically, the following aspects of communication are considered: (1) the ability of the nodes to multicast gossip messages, and (2) the size of the messages. The results describe the lower and upper bounds on the number of synchronous rounds required for the participants to discover each other. A particular question of interest is: if the network size is unknown, how does a node know that it has discovered all other nodes? Given a weakly-connected graph describing the initial knowledge of the nodes, every node in our algorithm can stop the discovery process knowing that there are no unknown nodes—this is done without any prior knowledge of the total number of nodes participating in the computation.     

Distributed decision propagation in mobile-agent proximity networks

This paper develops a distributed algorithm for decision/awareness propagation in mobile-agent networks. A time-dependent proximity network topology is adopted to represent a mobile-agent scenario. The agent-interaction policy formulated here is inspired from the recently developed language-measure theory. Analytical results related to convergence of statistical moments of agent states are derived and then validated by numerical simulation. The results show that a single (user-defined) parameter in the agent interaction policy can be identified to control the trade-off between Propagation Radius (i.e. how far a decision spreads from its source) and Localisation Gradient (i.e. the extent to which the spatial variations may affect localisation of the source) as well as the temporal convergence properties.

     

基于智能手机传感器数据的人类行为识别研究

运用智能手机传感器数据进行人类行为识别研究在医疗服务、智能环境和网络空间安全等领域有许多重要应用。目前,大多数的分类方法识别率都不高,尤其是在医疗服务领域。为了提高行为活动的识别准确率,先利用稀疏局部保持投影降维,将实验的数据集进行特征约简得到最优的实验特征子集,再用随机森林集成分类器完成了人类行为识别。实验结果证明,该方法不仅明显地降低了实验的特征数量,而且提高了整体精确度。     

Perimeter discovery in wireless sensor networks

In this paper, we focus on the perimeter detection problem using wireless sensor networks, as perimeter detection has a wide range of uses in several areas. We present a decentralized localized algorithm where sensor nodes determine if they are located along the perimeter of a wireless sensor network. Our proposed algorithm uses the location neighborhood information in conjunction with the Barycentric technique to determine if the sensor node enclosed by neighboring nodes, and consequently, if it is located within the interior of the wireless sensor network. We define performance metrics to analyze the performance of our approach and the simulation shows that the algorithm gives fairly accurate results.     

Asynchronous resource discovery in peer-to-peer networks

The resource discovery problem arises in the context of peer to peer (P2P) networks, where at any point of time a peer may be placed at or removed from any location over a general purpose network (e.g., an Internet site). A node (peer) can communicate with another node directly if and only if it knows a certain routing information to that other node. Hence, a critical task is for the peers to convey this routing information to each other.The problem was formalized by Harchol-Balter et al. and the specific P2P application they had in mind was the logical networks G₀ of servers placed by Akamai at various Internet sites. The routing information needed for a node to reach another peer is that peer’s identifier (e.g., IP address). A logical directed edge represents the fact that the peer at the tail of the edge knows the IP address of the one at its head. The problem is to compute the connected components in the underlying graph of G₀ (namely, the undirected graph obtained from G₀ by removing edge directionality). A number of algorithms were developed in Harchol-Balter et al. for this problem in the model of a synchronous network over a weakly connected directed graph. That is, the assumption was that some clock delivers pulses to every node every time unit at the same time, and a message delivery takes one time unit. The best of these algorithms was randomized. Subsequently, a deterministic algorithm for the problem on synchronous networks with improved complexity was presented in a paper by Kutten et al.The current paper extends the deterministic algorithm of Kutten et al. to the environment of asynchronous networks, where no clock pulses are assumed, and the message delivery time may vary and is not known. We managed to maintain complexities similar to those of the synchronous algorithm of Kutten et al. (translated to the asynchronous model). These are lower than the complexities that would be needed to synchronize the system. The main technical difficulty in a directed, weakly connected system is to ensure that nodes take steps that are consistent with each other, even if their knowledge about each other is not symmetric. Here, this task is further complicated by the fact that there is no timeout mechanism (which does exist in synchronous systems) to assist in ensuring consistency. In particular, as opposed to the case in synchronous systems, an asynchronous algorithm cannot first transform every directed edge to be bidirectional and second, apply an algorithm for bidirectional graph. Thus, our result takes another step towards representing the actual setting in a realistic manner.     

基于邻近服务的D2D节点发现技术综述

D2D节点发现技术是实现D2D通信首先要解决的重要问题。围绕无架构辅助的D2D节点搜索问题,对其研究现状进行了系统的梳理和述评,指出了该领域现有研究存在的问题,并综述了相应的典型解决方案和研究成果。     

On neighbor discovery in cognitive radio networks

A cognitive radio node is a radio device capable of operating over multiple channels. As a result, a network consisting of one or more cognitive radio nodes can adapt to varying channel availability in its geographical region by dynamically changing the channel (or channels) nodes are using for communication.     

Information discovery in mission-critical wireless sensor networks

Wireless sensor networks (WSN) are attractive for information gathering in large-scale data rich environments and can add value to mission-critical applications such as battlefield surveillance and reconnaissance. However, in order to fully exploit these networks for such applications, energy-efficient and scalable solutions for data storage and information discovery are essential. In this paper, we propose a novel method of information management in wireless sensor networks that can significantly increase network lifetime and minimize query processing delay resulting in quality of service (QoS) improvements that are of significant benefit to mission-critical applications. We propose novel methods for data dissemination, information storage and discovery that are totally distributed and also take into account the energy limitations of individual sensors. We present analytical and simulations results that prove the proposed methods of information management to offer significant improvements in the resolution of global ALL-type as well as individual ANY-type queries in comparison to current approaches. In addition, the results prove that the QoS improvements come with significant network-wide energy savings that will result in an increase of the network lifetime.     

Visual analysis of social networks in space and time using smartphone logs

We designed and applied interactive visualisation techniques for investigating how social networks are embedded in time and space, using data collected from smartphone logs. Our interest in spatial aspects of social networks is that they may reveal associations between participants missed by simply making contact through smartphone devices. Four linked and co-ordinated views of spatial, temporal, individual and social network aspects of the data, along with demographic and attitudinal variables, helped add context to the behaviours we observed. Using these techniques, we were able to characterise spatial and temporal aspects of participants’ social networks and suggest explanations for some of them. This provides some validation of our techniques.Unexpected deficiencies in the data that became apparent prompted us to evaluate the dataset in more detail. Contrary to what we expected, we found significant gaps in participant records, particularly in terms of location, a poorly connected sample of participants and asymmetries in reciprocal call logs. Although the data captured are of high quality, deficiencies such as these remain and are likely to have a significant impact on interpretations relating to spatial aspects of the social network. We argue that appropriately-designed interactive visualisation techniques–afforded by our flexible prototyping approach–are effective in identifying and characterising data inconsistencies. Such deficiencies are likely to exist in other similar datasets, and although the visual approaches we discuss for identifying data problems may not be scalable, the categories of problems we identify may be used to inform attempts to systematically account for errors in larger smartphone datasets.     

Knowledge discovery by inductive neural networks

A new neural network model for inducing symbolic knowledge from empirical data is presented. This model capitalizes on the fact that the certainty factor-based activation function can improve the network generalization performance from a limited amount of training data. The formal properties of the procedure for extracting symbolic knowledge from such a trained neural network are investigated. In the domain of molecular genetics, a case study demonstrated that the described learning system effectively discovered the prior domain knowledge with some degree of refinement. Also, in cross-validation experiments, the system outperformed C4.5, a commonly used rule learning system     

Hash-based proximity clustering for efficient load balancing in heterogeneous DHT networks

Distributed hash table (DHT) networks based on consistent hashing functions have an inherent load uneven distribution problem. The objective of DHT load balancing is to balance the workload of the network nodes in proportion to their capacity so as to eliminate traffic bottleneck. It is challenging because of the dynamism, proximity and heterogeneity natures of DHT networks and time-varying load characteristics.     

基于忠诚度的社交网络用户发现方法

针对社交网络中提高用户的高黏性问题,提出了一种基于用户忠诚度的用户发现的算法。该算法利用双重RFM模型对用户忠诚度进行计算,挖掘出忠诚度不同分类的用户。首先,通过双重RFM模型动态计算出用户在某一时间段的消费价值与行为价值,得到用户某一时间段的忠诚度;其次,根据用户的忠诚度,确定标度曲线,利用相似度计算找到典型的忠诚用户与不忠诚用户;最后,采用基于模块度的社区发现与独立级联传播模型,发现潜在的忠诚用户与不忠诚用户。在某社交网络的微博数据集上,实现了社会性网络服务（SNS）下用户忠诚度的量化表示,获得了基于用户忠诚度的用户发现结果。实验结果表明,所提算法能够有效挖掘出基于忠诚度的用户分类,可以为社交网站针对用户的个性化推荐及营销等,提供理论支持和实用方法。     

Infrared sensors using poly‐Si thin‐film transistors for proximity sensors integrated in smartphone displays

We have developed infrared sensors using poly‐Si thin‐film transistors (TFTs) for proximity sensors integrated in smartphone displays. Initially, we evaluate the infrared sensitivities of the poly‐Si TFTs, and it is found that a pin‐type TFT is suitable for the infrared sensors. Next, we propose three types of the infrared sensors. First, an analog current detection‐type sensor has a simple structure, and it is found that it can detect presence of a hand. Second, a lock‐in detection‐type sensor has tolerance against ambient light, and it is found that it can detect a target signal under noise signals. Third, a frequency detection‐type sensor has an advantage that only a digital circuit is necessary for detection, and it is found that it can detect the infrared intensity because the oscillation frequency increases monotonically with the intensity. We can utilize these infrared sensors on demand.     

人际信息交互中低功耗网络发现应用研究

无线人际信息交互网络围绕人际信息交互的特点展开。重点关注无线通信中的网络发现中的能耗问题,利用IEEE802.15.4网络的超帧结构,提出了一种通过变换超帧参数实现的低功耗网络发现算法,分析了该算法的能耗与时间约束的关系。最后,在硬件平台上开发了电子名片应用以测试网络发现算法的有效性。     

Cooperative device discovery for multi-interface self-organizing networks

For device-to-device communications, a device must locate neighbor devices on a certain channel within a short period to establish a communication link. When the channel information is not provided, a rendezvous technique can be adapted to resolve the issue of device discovery. Unlike existing approaches that have considered the rendezvous problem for a flat network and a single interface, the proposed approach considers a hierarchical network where nodes have different ranks and multiple interfaces. We extend the jump-and-stay rendezvous algorithm for multiple interfaces and divide the interfaces into two types: standby and scan interfaces. Scan interfaces follow the extended pattern of the jump-and-stay algorithm. Standby interfaces remain on a selected channel for a round of slots until rendezvous with other nodes. To determine the interface type of each node, we consider the rank of a node such that a higher-ranked node is assigned more standby interfaces. Based on this framework, we propose a cooperative rendezvous algorithm where a node rebroadcasts the channel information of a higher-ranked node together with its rendezvous message. Our simulation results confirm that the proposed rendezvous algorithm achieves superior performance compared to conventional methods.     

一种新的无线传感器网络拓扑发现算法*

GBGD是一种面向攻击的隐蔽性较强的拓扑发现算法,通过分析发现,该算法对实际网络进行了过于理想化的假设,导致无法在实际中应用。在GBGD算法工作模式的基础上,对实际网络提出了合理假设,设计实现了一种新的网络拓扑发现算法,通过对报文向基站汇聚过程中每一跳转发时延进行分析得出节点在路由树中的层次关系,进而推算出网络的拓扑。仿真实验结果表明,该算法能准确推断出网络的拓扑,并在报文存在丢失较多的情况下具有较好的鲁棒性。由Mica2节点组成的原型系统实验结果表明,该算法能够较好地应用于实际网络。     

Self-regularized causal structure discovery for trajectory-based networks

Trajectory-based networks exhibit strong heterogeneous patterns amid human behaviors. We propose a notion of causal time-varying dynamic Bayesian network (cTVDBN) to efficiently discover such patterns. While asymmetric kernels are used to make the model better adherence to causal principles, the variations of network connectivities are addressed by an adaptive over-fitting control. Compact regularization paths are obtained by approximate homotopy to make the solution tractable. In our experiments, cTVDBN structure discovery has successfully revealed the evolution of time-varying relationships in a ring road system, and provided insights for plausible road structure improvements from a traffic flow dataset.