• 期刊
• OpenAccess

Disaster Scenario and Record Capture System

Typical reports on past disasters contain valuable lessons that can help us improve our preparedness and response strategies and operations, especially with the help of modern analysis, simulation and visualization tools. However, disaster historical records are typically written in natural language for human consumption. They sometimes contain inconsistent data and information, which take time and effort to uncover. Extracting from the reports for further analysis by tools is almost impossible. DiSRC (Disaster Scenario and Record Capture) system is being designed as a solution of these problems. The system contains a cloud-based authoring system together with Disaster Record Capture Tools (DiReCT) running on mobile devices. Targeted users of DiReCT are government agents who are responsible for contributing data to be included in disasters records and volunteers trained by the responsible agencies. They can use DiReCT to capture observation data on the dynamics and effects of the disaster on places and people in machine readable form. The quality of the captured observation data is controlled and assured by both the capturing tools in real-time and by tools in the cloud. The authoring system can assist authors to write reports on the disaster based on the captured data and information. Disaster records produced with the help of DiSRC system can be read and processed by search, analysis, simulation and visualization tools for purposes such as developing better response strategies, providing decision support during emergencies, post-disaster analysis, and so on.

• 期刊
• OpenAccess

Ant Colony Optimization-based Resource Allocation and Resource Sharing Scheme for V2V Communication

The innovative architecture of Device-to-Device (D2D) underlying LTE/LTE-A networks is brought up to enable efficient discovery and communication between proximate devices. However, enabling D2D communications in a cellular network poses a major challenge that Quality of Service (QoS) requirements of D2D communications need to be guaranteed. Thus, synchronization between devices becomes a necessity and Radio Resource Management (RRM) becomes a key design aspect to enable D2D communication, where resource allocation phase is one of the most critical aspects. The problem of resource allocation in D2D communication system is a combinatorial optimization issue, difficult to obtain optimum solutions in polynomial time. In order to reduce complexity, it can be solved by using linear algorithms or by metaheuristic methods. In this paper an Ant Colony Optimization (ACO) based resource allocation and resource sharing scheme for Vehicle-to-Vehicle (V2V) based D2D communications in LTE-A networks is introduced. The swarm intelligence algorithm ACO, which is a typical algorithm of metaheuristic methods, is adopted to resolve the optimization problem of maximizing the network sum rate while considering the QoS requirements.

• 期刊
• OpenAccess

Unmanned Aerial Vehicle based Reliable and Energy Efficient Data Collection from Red Alerted Area using Wireless Sensor Networks with IoT

Most important characteristics of the Wireless Sensor Networks (WSNs) is self-configuration, in which each deployed sensor node configures itself with neighbor node and establishes the network topology. In the proposed work, an Unmanned Aerial Vehicle (UAV) based Reliable and Energy Efficient Data Collection Mechanism (REEDCM) is introduced using Wireless Sensor Networks with the Internet of Things (IoT). It mainly focuses on the effective data collection from the man un-attended area (i.e., red alerted area) including nuclear disaster zone, volcanic eruption, forest fire, and battlefield, etc. Battery energy is a major constraint of the WSNs, thus an efficient data collection mechanism is needed to enhance the energy efficiency and scalability. In REEDCM, a novel clustering mechanism is presented which selects the Cluster Head (CH) based on the residual energy, speed and number of neighbors. Besides, the UAV Data Collector (UAV_DC) is used to gather the data directly from the CHs which in turn transmits the collected data to the Base Station (BS), then the BS shares the sensor information to the users through the Internet. From the simulation results, it is revealed that the proposed REEDCM provides better network performances in terms of average residual energy, average end-to-end delay, Packet Delivery Ratio (PDR) and throughput.

• 期刊
• OpenAccess

Smartphone Holding Styles Based Step Detection and Length Estimation

In this study, we propose an effective method for accurately detecting the number of walking steps and estimating the step length adaptively using the set of inertial sensors of a smartphone. The proposed walking behavior recognition method can be used as an important functional block in a pedestrian dead reckoning system. We develop a method for classifying the four main holding styles while walking, i.e., holding a phone in the hand while watching it, holding a phone while calling, swinging it, and putting it in a pocket. The four main holding styles are divided into 34 sub-styles, which encompass the various free styles of holding a smartphone during daily activities. Using this holding style classification, we obtain better performance when counting the walking steps and estimating the step length, although we only employ a set of feature values that are easily calculated without any complex data processing techniques. Based on numerous experiments, we demonstrate the excellent performance of the proposed method for step counting and step length estimation for various holding styles.

• 期刊
• OpenAccess

Automatic Text Recognition in Natural Scene Using Neural Network Classifier with Dynamic-group-based Hybrid Particle Swarm Optimization

This paper presents a two-stage algorithm for automatic text detection and recognition. In the first stage, using a stroke width transform and an improved connected component, an edge analysis method detects a candidate character region. Subsequently, a text region is located by filtering and linking characters with similar font sizes and colors. For the second stage, a histogram of oriented gradient is employed as a feature descriptor, and a neural network classifier is built with dynamic-group-based hybrid particle swarm optimization (DGHPSO) for character recognition. In DGHPSO, each group's threshold value of similarity depends on the threshold values of fitness and distance. In addition, a local search algorithm is used to improve the search for a global optimum. The proposed algorithm was experimentally validated; it outperformed a number of recently published studies in terms of the text recognition rate when tested on the ICDAR 2003 database and the Street View Text database.

• 期刊
• OpenAccess

Regression Analysis of Coherence between Concurrent EEG-EEG and EEG - CPPG Signals from Prefrontal Cortex During Music Evoked Emotions

This paper summarizes the results of experiment that focus on effect of music evoked emotions on mean squared coherence and phase coherence between two signals with same nature (electrical brain signal - EEG) and the two signals with different nature (EEG and hemodynamic brain signal CPPG). Regression analysis was carried out to find out the mathematical relation between mean squared coherence and phase coherence estimated between the two signals captured from prefrontal cortex and the physiological parameters. Number of synaptic connections between the two measurement sites/ signals and its strength is reflected in the coherence. It is a quantitative measure of association between the two simultaneously acquired signals as a function of frequency. Physiological parameters studied by the authors are SBP, DBP, HR, Blood Glucose and BMI. These are some of the potential biological markers closely related to emotional response. Data was collected from twenty multi-lingual subjects of both the genders with Mean_(age) = 39.25 years and SD_(age) = 11.625 years. No strong correlation was found between the coherence (calculated between EEG-EEG and EEG-CPPG) and the physiological parameters studied during the various emotional states. It was observed that MS Coherence between the signals with similar nature is higher than it in between the signals with dissimilar nature. T-paired test was carried out to show that the means of these two coherences is different. Coherence between EEG-EEG was compared with coherence between EEG-CPPG and it was observed that they are very different (p ＜ 0.001), were as when coherence between EEG-EEG (or EEG-CPPG) is compared with coherence between EEG-EEG (or EEG-CPPG) p-value was ＞ 0.05. This technique can be applied to wider population in the field of clinical neuroscience with or without any known neurological disorder. More connectivity measures can also be included to study music evoked emotions or stroop task.

• 期刊
• OpenAccess

Combining Mutual Information and Entropy for Unknown Word Extraction from Multilingual Code-Switching Sentences

In multilingual environments, a single statement may include content from more than one language, a phenomenon known as code-switching. Among speakers of Mandarin Chinese, code switching is a frequent occurrence in daily life, and this mixing of different languages poses serious challenges for language processing. This paper collects text corpora including code switching between Mandarin and English and Mandarin and Taiwanese, where Mandarin is the dominant language. Mutual information and entropy are then used as a basis for an algorithm to identify unknown words from multilingual texts which are then automatically referenced for multilingual inclusions. Experimental results show that the proposed method effectively filters unrelated new words, thus improving the accuracy of extracting unknown words.

• 期刊
• OpenAccess

Breaking Text-Based CAPTCHAs using Average Vertical Partition

CAPTCHA, which stands for Completely Automated Public Turing Test to Tell Computers and Humans Apart, has been widely used as a security mechanism to defend against automated registration, spam and malicious bot programs. There have been many successful attacks on CAPTCHAs deployed by popular websites, e.g., Google, Yahoo!, and Microsoft. However, most of these methods are ad hoc, and they have lost efficacy with the evolution of CAPTCHA. In this paper, we propose a simple but effective attack on text-based CAPTCHA that uses machine learning to solve the segmentation and recognition problems simultaneously. The method first divides a CAPTCHA image into average blocks and attempts to combine adjacent blocks to form individual characters. A modified K-Nearest Neighbor (KNN) engine is used to recognize these combinations, and using a Dynamic Programming (DP) graph search algorithm, the most likely combinations are selected as the final result. We tested our attack on the popular CAPTCHAs deployed by the top 20 Alexa ranked websites. The success rates range from 5.0% to 74.0%, illustrating the effectiveness and universality of our method. We also tested the applicability of our method on three well-known CAPTCHA schemes. Our attack casts serious doubt on the security of existing text-based CAPTCHAs; therefore, guidelines for designing better text-based CAPTCHAs are discussed at the end of this paper.

• 期刊
• OpenAccess

Secure Outsourcing Algorithm for Signature Generation in Privacy-Preserving Public Cloud Storage Auditing

Public cloud storage auditing allows a file owner or a public verifier to conduct integrity checking without downloading the whole file from cloud server. Plenty of unaffordable modular exponentiations for resource-constraint devices are required on the client side in the process of signature generation for public cloud storage auditing. In this paper, we introduce a secure outsourcing algorithm to make cloud storage public auditing more feasible and practical for the computationally limited clients. Compared with the existing algorithms for secure outsourcing of modular exponentiations, the proposed algorithm designed for this scenario takes advantage of that the bases of modular exponentiations do not need to be protected. Therefore, our algorithm is efficient. Furthermore, the evaluation shows that it has high efficiency and checkability as well. Specifically, the client-side cost is reduced to 5.6% of the original computation cost, and the probability that the client detects the misbehavior of server is close to 1.

• 期刊
• OpenAccess

Sentence-Ranking-Enhanced Keywords Extraction from Chinese Patents

Patent keywords, a high-level topic representation of patents, hold an important position in many patent-oriented mining tasks, such as classification, retrieval and translation. However, there are few studies concentrated on keywords extraction for patents in current stage, and neither exist human-annotated gold standard datasets, especially for Chinese patents. This paper introduces a new human-annotated Chinese patent dataset and proposes a sentence-ranking based Term Frequency-Inverse Document Frequency (SR based TF-IDF) algorithm for patent keywords extraction, motivated by the thought of ＂the keywords are in the key sentences＂. In the algorithm, a sentence-ranking model is constructed to filter top-K_S percent sentences from each patent based on a sentence semantic graph and heuristic rules. At last, the proposed algorithm is evaluated with TF-IDF, TextRank, word2vec weighted TextRank and Patent Keyword Extraction Algorithm (PKEA) on the homemade Chinese patent dataset and several standard benchmark datasets. The experimental results testify that our proposed algorithm effectively improves the performance of extracting keywords from Chinese patents.