Smartphone Data Protection Using Mobile Usage Pattern Matching

Introduction

Smartphones are extremely popular and convenient these days. People carry them anytime, anywhere and use them to perform daily activities like making phone calls, checking emails, and browsing the mobile Web. However, they are easily lost or stolen because of their small sizes and high mobility. Personal data such as addresses and messages stored in the devices are revealed when the devices are lost (Ghosh & Swaminatha, 2001). Various methods have been used or proposed for smartphone data protection. The methods can be classified into five categories: (i) password/keyword identification, (ii) human intervention, (iii) biometric-based identification, (iv) anomaly/behavior-based identification, and (v) other ad hoc methods. They will be introduced in the next section. This research, an anomaly/behavior-based identification, applies the methods of handheld/mobile usage data matching to smartphone data protection. The idea is to have devices identify the user, analyze usage pattern, and take any necessary actions to protect confidentiality of sensitive data stored. More specifically, we would like to compare statistical anomaly-based or behavior-based user pattern sample activities with typical usage profile known to be normal. To our best knowledge, there is no research work in this area by applying pattern recognition techniques. Experiment results show that this method is effective and easy-to-use.

This research is to design and implement a strategy of smartphone data protection, which includes the following features in order of importance:

• •

  Rigorous and effective handheld data protection: It is the major objective of this research.

• •

  Easy to use and apply: Many security methods are abandoned because the users are reluctant to learn how to use them.

• •

  Easy to adapt to each individual owner: When the device owner is changed, the methods can quickly, easily adapt to the new owner.

A set of novel approaches is proposed in this chapter to protect handheld data by using usage pattern identification. The proposed methods are divided into five steps:

• 1.

  Usage data gathering, which is to collect the device usage data,

• 2.

  Usage data preparation, which removes noises from the raw usage data,

• 3.

  Usage pattern discovery, which finds valuable patterns from the prepared usage data,

• 4.

  Usage pattern analysis and visualization, which is to analyze and display the discovered patterns for finding hidden knowledge, and

• 5.

  Usage pattern applications, one of which is smartphone data protection used in this research.

This chapter is organized as follows. A background study consisting of three major subjects is given in the next section. Section III introduces our proposed system using handheld usage pattern matching. Two algorithms of structure similarity are used to check against any possible unauthorized uses: (i) approximate usage string matching and (ii) usage finite automata. The two methods are explained in the following two sections. Section VI shows and discusses some experimental results. Conclusion and some future directions are given in the last section.

Background

This research includes three themes:

• •

  Mobile handheld computing, which is the computing for (smart) cellular phones.

• •

  Handheld security, which detects abnormal handheld data accesses and protects the data from unauthorized uses.

• •

  Approximate string matching, which is to find the “best” match of a string among many strings.

Related research of these themes will be discussed in this section.