Biometric Gait Authentication for Smart Phone on Hand Paper Example

Introduction

The world continues to be ever reliant on smart devices particularly on smartphones. Typical applications include communications, personal healthcare transactions, and entertainment (Ahn and Han 3, Derawi, Nickel, and Bours 1). Therefore, it implies that personal information continues to be collected and analyzed through smart devices. The increased exposure of personal information poses security and privacy concerns which the current authentications system has not been able to address comprehensively. Typical forms of authentications employed in smart devices include biometric recognition such as the fingerprints, gait and face recognition, password and patterns (Birgale and Kokare 448).

Is your time best spent reading someone else’s essay? Get a 100% original essay FROM A CERTIFIED WRITER!

Save 25%

In comparison, biometric authentication is of more convenience to the user in place of tokens and passwords. Despite the said advantage biometric authentication usually utilizes machine learning and pattern recognition in their systems to cope with natural changes in biometric measurements (Birgale and Kokare 448). The original biometric template is kept under unconcealed format to match with biometric samples when identifying the user. Holding in unconcealed form could potentially expose the critical security systems leaving the user privacy vulnerable when used on smartphones. Since smartphones are easily misplaced, someone could gain access to the original biometric template by accessing the phone repository. If the smartphone is also integrated with the financial information of the user, it places the financial information at risk.

Gait recognition as one of the biometric authentication systems promises human identification. Due to technological advancement, human motion analysis has been integrated with biometrics for securing sensitive facilities such as airports, military installations and banks (Rani and Arumugam 1). Rani illustrates that gait recognition is based on the fact that each has a specific walking style and thus can be employed in human recognition. Gait recognition termed as a biometric means which facilitates an automatic verification of an individual based on the walking style (Derawi, Nickel and Bours 2). Derawi outlines that the benefit of using gait recognition is that it is an unobtrusive authentication method which is convenient to the user. An essential procedure in gait authentication and identification is extracting gait input signals (Gupta et al. 379).

Background and Statement of the Problem

Authentication is termed as the means of establishing personal identity and whether the status is as the person claims to be (Gupta et al. 378). With the increasing interdependence of smart devices, authentications form security aspects. Various forms of authentications are in use which is either based on biometrics (including gait, face, and fingerprint), tokens or passwords. Biometric authentication has emerged as a useful form of securing smart devices. Biometric identifications utilize physiological features of human beings for personal identifications (Li, Jiang, Jia, and Lin 823). With the development of gait recognition as a means of personal identifications, it presents a revolutionary technology in biometric identifications (Li et al. 83). Gait recognition presents a lot of benefits including avoidance of disguise, minimal resolution requirements, and being noninvasive (Li et al. 823). Various methods are employed to obtain gait recognition data which consists of video sensors, Kinect based camera, a wearable sensor, and tactile sensor-based method. In biometric authentications, systems extractions of the gait features are employed to encrypt biometrically,

This project proposal put forward authentication systems reliant on biometric gait recognition with a bid to improve on user's privacy and systems security on smartphones. The intention is to employ biometric model reliant on human gait. In essence, it intends to make use of inertial sensors such as a wearable accelerometer. Accelerometer sensors have been used in different applications on smart devices to greater success (Hong et al. 331). The system will utilize committed schemes where the key will the authentication determinant, encrypted biometrically using the personal gait of the user. Data on the gait features will be collected using a smartphone held by the user. Gait features will be extracted and analyzed in real time. Efficient identification rate will be achieved by employing SVM classifier and PCA feature reduction technique. The systems will have considerable benefits with regards to space utilization and minimal computational needs. Through this, it emphasizes the need for applications directly to smartphones which have small storage in comparison with systems such as PR-ML (Hong et al. 328).

There is an increased reliance on smartphones fueled by technological growth. Smartphones have used in various industries such as transactions, information, entertainment, communications, and personal health continue to dominate human lives (Ahn and Han, 3), (Derawi, Nickel, and Bours 1). In this regard, there is an increased collection of the personal database which continues to be collected and analyzed. The application poses security risks to private information which might spill onto financial services. The convectional authentication systems utilized in smartphones utilize biometric, token and password recognition to secure the devices. However, these methods have been unable to deal with security concerns comprehensively. The primary forms of biometric authentication in handheld devices have been mainly being fingerprints (Birgale and Kokare 448).

As such, a more secure way of authentication is required to ensure the private information of the user. Biometric gait authentication promises to be a more secure technology of authentication since humans possess unique gait traits. The method has advantages over the traditional authentication systems concerning space utilization and user convenience (Gupta et al. 378). However, employing biometric authentication on smartphones presents a security threat in case someone obtains the original biometric templates. Gaining original biometric templates could leave the users biometrics exposed, and as results, it would require forfeiture of the device or risk losing sensitive data. To achieve this, it calls for a secure gait biometric authentication system which encrypts the biometrics data. Wearable gait recognition sensor offers an alternative to gait recognition. It can be utilized to retrieve data on personal identity by using external sensors.

Research Questions, Aims, and Objectives of the Research

Aims and Objectives

The objective of this research project is to develop gait authentication systems reliant on biometric gait recognition to improve on systems privacy and security for smartphones. The proposed model aims to utilize a wearable accelerometer sensor to obtain gait data from the user. Using a committed scheme, the model will secure the device through biometric authentication encrypted from the gait of the user. The gait data will be used to obtain the encrypted password after which it will be discarded to guarantee the privacy and security of the user. Thus, novel research will be conducted to facilitate the secure integration of gait authentication systems with smartphones. The project aims to develop a biometric gait recognition with a high identification rate through the use of

Research Questions

This project will seek to address the following issues:

• What are the methods which can be utilized for analysis of the gait data for biometric identifications and authentications?
• How can the gait recognition data be obtained and stored on smartphones and carry out authentications for the devices?
• How can the optimum identification rate for the system be achieved?
• How can the biometric templates be secured to eliminate the vulnerability of the authentication system?

Integrating biometric authentication on smartphones will enhance their privacy and security.

Data Collection Methods and Instruments

To obtain gait signals of the user a Samsung smartphone held by the user will be used. Wearable sensors can be attached to various body parts (Li et al. 1). The discrete time signals are as a result of inertial and gravity acceleration as well as ground reaction forces. The smartphone will be attached to the user and oriented horizontally. The gait signals will be captured using the inbuilt three-dimensional accelerometer sensor while the user walks. The Samsung smartphone utilizes an Android platform which has inbuilt software for accessing the accelerometer. The sensor output data will be sent to a file by use of the written software. The assumed samples per second for all x, y, and z-direction will be about 40-50.

Integration of smartphones and accelerometer sensors reduces the sampling rate of the accelerometer as it depends on the smartphone OS. The frequency of two successive return samples is inconsistent. Thus, to get a consistent output signal the obtained signal will have to be interpolated by employing linear interpolations. This will result in fixed frequency while ensuring a fixed time interval for two successive samples emanating from the same point.

Methodology

Gait features will be extracted and analyzed in real time. Efficient identification rate will be achieved by employing SVM classifier and PCA feature reduction technique. The input data to the proposed system will be the gait video sequences obtained by the inbuilt accelerometer. Use of the background subtraction will develop binary silhouettes. This will involve two processes namely: foreground extraction and background modeling. Numerical information will then be obtained from the extracted features. The information will be stored in a database and will be used when developing the pattern recognition classifiers. Background modeling will utilize three techniques including histogram based background modeling, use of median values and change detection mask technique. The project proposes to utilize a BCS with a fuzzy commitment scheme (Wanare 1).

For the enrollment phases, gate signals of the user Q will be obtained and preprocessed to minimize influences from the environment obtained. Vector features will be extracted in both frequency and time domains after which they will be binarized. Then, dependable binary feature k will be obtained, and it will be a factor of determining reliable components. Simultaneously, cryptographic key S will be developed in a random manner and equivalent to a specific user will be encoded for codeword C by utilizing correction codes for the errors.

Computation of the secured and hash value n will be carried out by a committed fuzzy scheme F with the use of binding function and hash value respectively. The assistance data are to be employed in extracting the dependable binary features are to be stored locally for application during authentications. For authentication, the users will give various samples of their gaits. Utilizing the assistance data, the under storage from the enrollment phase, a reliable vector will be retrieved while the gaits will be preprocessed to obtain a features vector. The computation of a corrupted codeword c with and binding w will be carried out by a decoding function f after which an equivalent error correction decoding algorithm does a retrieval of cryptographic key m from c. Consequently, the 'm' hash value is to be compared with h (n) for authentication.

Signal Preprocessing and Feature Exploration Classification

Noise filtering

As the accelerometer sensor will be sampling user's movement, noise data will potentially be collected. Smartphone mounted accelerometer generates high noises in comparison with a standalone sensor as its operation is r...