Chao Tong

Complex networks properties analysis for mobile ad hoc networks

Recently, research on complex network theory and applications draws a lot of attention in both academy and industry. In mobile ad hoc networks (MANETs) area of research, a critical issue is to design the most effective topology for given problems. It is natural and significant to consider complex networks topology when optimising the MANET topology. Current works usually transform MANET or sensor network topologies into either small-world or scale-free. However, some fundamental problems remain unsolved. Specifically, what are the average shortest path length, degree distribution and clustering characteristics of MANETs? Do MANETs have small-world effect and scale-free property? In this work, the authors introduce complex networks theory into the context of MANET topology and study complex network properties of the MANETs to answer the above questions. The authors have theoretically analysed the degree distribution and clustering coefficient of MANETs and proposed approach to computing them. The degree distribution and clustering coefficient of MANETs are theoretically deduced from node space probability distribution on different mobility models (including but not limited to random waypoint model). Simulation results on average shortest path length, clustering coefficient and degree distribution show that in most cases MANETs do not have the small-world effect and scale-free property.

A novel green algorithm for sampling complex networks

Researches of complex networks such as social networks are becoming popular in recent years. Due to the large scale and complex structure of these networks, analysis and studies on a complete network require a lot of computational resources and storage space, which will also consume a large amount of energy. Sampling algorithms provide a new green approach for this problem. Especially some researches related to network communities with high energy consumption can be directly conducted on the sampled networks, which maintain the community structure of original networks. In this paper, we propose a sampling algorithm named Improved Forest Fire Sampling algorithm based on PageRank (IFFST-PR) based on the idea of Forest Fire Sampling and PageRank algorithm. IFFST-PR can maintain the community structure of original networks. We select a set of key nodes called community cluster center, according to a coefficient named community coefficient. Besides, we adopt PageRank to decide the order of initiative sampling nodes. To make a comprehensive comparison of IFFST-PR with other 6 algorithms, we use network community profile and Kolmogorov-Smirno D statistics to prove the consistency between sampled networks and original networks. Experiments applied on 3 different data sets show that IFFST-PR has better performance in terms of most parameters defined in network community profile than those of the other 6 algorithms. HighlightsA sampling algorithm which can maintain community structure is proposed.Sample a network by the node order generated by the PageRank algorithm.Start sampling from marginal nodes to protect the structure of small communities.Start sampling from key nodes of networks called community cluster center.Using network community profile (NCP) to comprehensively evaluate the algorithm.

An efficient deep model for day-ahead electricity load forecasting with stacked denoising auto-encoders

Abstract In real word it is quite meaningful to forecast the day-ahead electricity load for an area, which is beneficial to reduction of electricity waste and rational arrangement of electric generator units. The deployment of various sensors strongly pushes this forecasting research into a “big data” era for a huge amount of information has been accumulated. Meanwhile the prosperous development of deep learning (DL) theory provides powerful tools to handle massive data and often outperforms conventional machine learning methods in many traditional fields. Inspired by these, we propose a deep learning based model which firstly refines features by stacked denoising auto-encoders (SDAs) from history electricity load data and related temperature parameters, subsequently trains a support vector regression (SVR) model to forecast the day-ahead total electricity load. The most significant contribution of this heterogeneous deep model is that the abstract features extracted by SADs from original electricity load data are proven to describe and forecast the load tendency more accurately with lower errors. We evaluate this proposed model by comparing with plain SVR and artificial neural networks (ANNs) models, and the experimental results validate its performance improvements.

A convolutional neural network based method for event classification in event-driven multi-sensor network

We investigate the similarities between an actual image and sensor data.We propose a CNN-based method to improve the event classification accuracy.An variant of AlexNet is designed and established for classifying.The results indicate this CNN-based classifier outperforms than kNN and SVM methods. Display Omitted A multi-sensor network usually produces a large scale of data, some of which represent specific meaningful events. For event-driven multi-sensor networks, event classification is the basis of subsequent high-level decisions and controls. However, the accuracy improvement of classification is always a challenge. Recently the deep learning methods have achieved vast success in many conventional fields, and one of the most popular deep architectures is convolutional neural network (CNN) which sufficiently utilizes partial features of the input images. In this paper, we make some analogy between an image and sensor data, then propose a CNN-based method to improve the event classification accuracy for homogenous multi-sensor networks. An variant of AlexNet has been designed and established for classifying the event by acoustic signals. The results indicate that this CNN-based classifier outperforms than k Nearest Neighbor (kNN) and Support Vector Machine (SVM) methods on our data set with a higher accuracy.

Effects of Some Lattice Reductions on the Success Probability of the Zero-Forcing Decoder

Zero-forcing (ZF) decoder is a commonly used approximation solution of the integer least squares problem, which arises in communications and many other applications. Numerical simulations have shown that the LLL reduction can usually improve the success probability PZF of the ZF decoder. In this letter, we first rigorously show that both SQRD and V-BLAST, two commonly used lattice reductions, have no effect on PZF. Then, we show that LLL reduction can improve PZF when n = 2, we also analyze how the parameter δ in the LLL reduction affects the enhancement of PZF. Finally, an example is given which shows that the LLL reduction decrease PZF when n ≥ 3.

Stability of Exact and Discrete Energy for Non-Fickian Reaction-Diffusion Equations with a Variable Delay

This paper is concerned with the stability of non-Fickian reaction-diffusion equations with a variable delay. It is shown that the perturbation of the energy function of the continuous problems decays exponentially, which provides a more accurate and convenient way to express the rate of decay of energy. Then, we prove that the proposed numerical methods are sufficient to preserve energy stability of the continuous problems. We end the paper with some numerical experiments on a biological model to confirm the theoretical results.

How Online Social Network Affects Offline Events: A Case Study on Douban

Social networking sites (SNS) such as Facebook and Twitter are becoming popular forms for finding, promoting and attending offline events and activities. Much work has looked into characterizing these social networks and their user behavior. With the emergence of event and activity-based applications such as Facebook Events, Linked In Events, Xinghui, Zaizher, and Douban, it is easier to connect with other people online and meet them offline. However, few have looked into analyzing these offline events and activities that are shared online. This paper seeks to gain insights into the user behavior around people attending offline events which are promoted online. By studying the events in Douban, we present results around the event properties, user behavior of participants and wishers to an event, and social influence to an event. We show that event distribution by participants and wishers follow the typical power-law distribution, most users attend or like short events that last several days or regular events that last less than 3 months, participants attend an event within one day after the publish time, and that there is an exponential relationship between follow probability and number of common events attended between two users and a linear relationship for common events interested in. These findings provide a better understanding on how SNS could affect user behavior in attending events, and provide guidelines on how to improve the design of event-based applications.

Real-time generation of personalized home video summaries on mobile devices

Abstract With the proliferation of mobile devices and multimedia, videos have become an indispensable part of life-logs for personal experiences. In this paper, we present a real-time and interactive mobile application for home video summarization on mobile devices. The main challenge of this method is lack of information about the video content in the following frames, which we term “partial-context” in this paper. First of all, real-time segmentation algorithm based on partial-context is applied to decompose the captured video into segments in line with the change in dominant camera motion. Secondly, an original key frame update strategy is presented to optimize selected key frames in such partial-context. In addition, the main challenge to conventional video summarization is the semantic understanding of the video content. Thus, we leverage the fact that it is easy to get user input on a mobile device and attack this problem through the user interaction. The user preference is learned and modeled by a Gaussian Mixture Model (GMM) whose parameters are updated each time when users manually select a key frame. Our system utilizes the user preferences to optimize the key frame update process. Evaluation results demonstrate that our system significantly improves users experience and provides an efficient automatic/semi-automatic video summarization solution for mobile users.

Evolution of disconnected components in social networks: Patterns and a generative model

The majority of previous studies have focused on the analyses of an entire graph (network) or the giant connected component in a graph. Here we study the disconnected components (non-giant connected components) in real social networks, and reporting some interesting discoveries on how these disconnected components evolve over time. We study six diverse, real networks (citation networks, online social networks, academic collaboration networks, and others), and make the following major contributions: (a) we make empirical observations of the longevity distribution of disconnected components, and find that the curve of the distribution demonstrates a decaying trend; (b) we find that the distributions of final size of disconnected components that merge with one another or get absorbed by the giant connected component both follow power laws; (c) we find that the majority of mergings are between disconnected components and the giant connected component. The mergings that happen among disconnected components are small in scale (involve only a few components). The longevity distributions of the disconnected components in those mergings are similar, where the shortest-lived disconnected components are the most in number; and (d) we propose an empirical generative model that can produce the networks with our observed patterns.

Detecting overlapping communities of weighted networks by central figure algorithm

In recent years, the community structures in complex networks has become a research hotspot. In this paper, we focus on weighted networks and propose a unique algorithm on detecting overlapping communities of weighted networks based on central figure with considerable accuracy. In the algorithm, all the central figures are first extracted. Then to each central figure, nodes are absorbed by closures and weak ties. The experiments are based on LFR Benchmark. Through the experiment, we can know that the performance of our algorithm is better than that of COPRA (Community Overlap Propagation Algorithm) algorithm.