A Hidden Markov Model and Internet of Things Hybrid Based Smart Women Safety

A Hidden Markov Model and Internet of Things Hybrid Based Smart Women Safety 

Abstract—Smart technologies for women safety are gaining popularity over the last few decades. Several nefarious approaches to women that outraged the entire nation awakened the scientists globally to design smart apps for women safety. This paper proposes a concept of a multivariate security paradigm for women under possible offensive threats by deep sensing approaches. Internet of Things (IOT) based platform provides dexterity and dynamicity in correlating a plethora of sensors and actuators to ensure women safety. Hidden Markov Models (HMM) offer scope for better predictive analysis for its dynamic probabilistic nature and helped us developing a dense sensing approach based on traces of suspicious activities. There is a situation-based analysis for relative modelling based on face recognition as well as fuzzy labeling of verbal conversations. https://codeshoppy.com/shop/product/women-security-app/ If an emergency situation is triggered, a GSM/GPS module will generate emergency in case of after-shock otherwise will warn the female device bearer. The results of experimentations proved to be really promising with an accuracy of 94.7%.Keywords-Internet of Things, Hidden Markov Models, Smart Women Safety, GSM/GPS module, fuzzy.

INTRODUCTION 

The concept of equality in a global society often faces a challenge when it comes to female population for unavoidable male dominated behaviors. Various circumstances have triggered the necessity to ensure women security and protection. According to the statistics of National Crime Records Bureau, the highest rate of women harassment increased by 65% over last two years [1] and reached 37% in the last decade. Most of the existing researches have complex alert systems which are to be activated by some input from the victim, like, pressing an SOS button [2], [3] or shouts [4]. Practically, often the victims cannot realize they are befooled by an unscrupulous company or when they are under threat, they are not left with enough chances to hit a button or shout in the top of their lungs. Internet of Things (IOT) based researches gained popularity over the last few decades for offering an agile paradigm to manage quite a large number of devices from a singular master control [5]. Still, most of the works are restricted in designing fancy luxurious smart residences [6]. 

On the other hand, Hidden Markov Model is a hotcake nowadays but offer limited applications in gesture [7] and handwriting recognitions [8]. Recently, a work on adaptive learning based IOT-HMM hybrid oriented smart healthcare gained popularity based on its vast scope of dynamic applications [9]. Large scale wearables are developed recently [10] but there is a major problem of synchronicity in the discrete modules of the children safety paradigm. Biometrics based identification of danger is a better option [11] but it has overlooked the common flaw of practical applications and mishandling of the scheme can lead criminals to mimic the patterns and harass their target quite easily. This paper proposes a multivariate security paradigm to protect female population from possible criminal activities which may affect their security and safety. HMM can be simply understood as a doubly stochastic Bayesian Network [12] which provides a deep sensing approach for predicting possible insecurities based on minimal training data. A HMM-IOT hybrid triggers the predictive and analytical modules of the security paradigm. A plethora of sensors are logically managed by the IOT platform and a scheme of pre-shock and after-shock is developed. 

If an acquittance plans or decides to make nefarious approach the HMM module plugs in and alerts the female detector bearer. If there is a sudden approach in an unexpected scenario a forty seconds fast track prediction mode of the HMM-IOT hybrid turns on the emergency module and the GSM/GPS couple updates location and activates emergency request to pre-assigned contacts, medics and nearest police stations. Rest of the paper is organized as follows.