Enhancing Smart Home Security with AI and Machine Learning Tech

This report discusses the application of machine learning in smart home security systems to overcome the limitations of traditional security measures. The proposed system integrates signal equipment, sensors, video cameras, controllers, and security system software to create a multi-layered protection system. It utilizes face detection, pattern recognition, and neural networks to identify intruders and differentiate them from residents. The system's algorithm involves sensor activation, video recording, face detection using OpenCV and clmtrackr, and automated decision-making based on neural network analysis. The neural network is trained with data collected during normal activity patterns to identify anomalies. The report highlights the system's usefulness in providing real-time alerts and enhancing the safety of vulnerable individuals. Future modifications include cost reduction and incorporating user feedback into the decision-making process. The system aims to provide an intelligent and effective solution for home security, addressing the drawbacks of previous systems and ensuring resident safety.

Running head: SMART SECURITY
SMART SECURITY
Name of the Student:
Name of the University:
Author Note:

1SMART SECURITY
The advancement of the technology enables the use of machine learning in various
purpose. One of the interesting uses of the machine learning is to ensure the security and
implementation of smart home. It has been found out that the typical systems are sometimes
failed to detect the tricks of the buglers. The security system in the smart house will be based on
the machine learning, so that it will be intelligent enough to detect the advanced tricks of the
buglers.
Discussion
Problem: The problem of the current security system of the house security is that it
cannot always prevent the security threats.
Retrospective solution of the problem: In the past the, these kind of problems were solved
by the improvement of the sensor components and the advanced alarming system . However,
these steps were helpful at initial stage of implementation but it fails to protect the security in
long run as the intruders are also using the advanced technology for the trespassing.
Smart house security system: In order to solve this problem , machine learning , which is an
integrated part of artificial intelligence is used (Artem & Vasyl,2017). The whole system based
on the machine learning process makes the security system smart. The components of the
security system in the smart house are-
 Signal equipment
 Sensor
 Video Camera
 Controller
 Security system software

2SMART SECURITY
Structure of the proposed security system
(Source: Created by the author)
Each of the equipment has different functions in maintaining the security of the house. The
functions of the different components are triggered on the basis of the situation happening with
the security system in the house.
There are different levels of protections of the security. The first level of the protection is the
perimeter protection which is done by the monitoring camera.
In case, if the intruders can still manage to get into the house they will be caught by the
monitoring cameras and sensors installed into the house.
The third level of the security may include the hidden cameras and the sensors that responds
on the human temperature.
The forth level of the security helps to detect the intruders using the machine learning
mechanism. The further level of the security is activated in case the intruder can manage to
bypass all the three security levels of the system.

3SMART SECURITY
The face detection is also used in maintaining the security of the house. The pattern detection of
the machine learning can be used in this case (Jacobsson, Boldt & Carlsson,2016). The system
can recognize the face pattern of the owner of the hose and can differentiate it with the pattern of
the intruders.
Implementation of the Artificial Intelligence in the system:

4SMART SECURITY
Figure 1: Flow chart of the working principal of the proposed system
(Source: Artem & Vasyl,2017)
Algorithm of the proposed smart security system:
The network sensor used in the system triggers when it feels the movement of the human
in the house. Initially, the video recording and video streaming turned off to save the energy. In
case, the sensor is triggered the video camera starts to work automatically. Cameras are turned
for the face detection. In case if the system founds the face pattern matching with the pattern
recorded n the system it will not set the alarm on. Unless it will set the alarm on and will inform
the owner of the house through SMS.
There proposed smart security system is made with some components which have different
functionality.
Recognition of face:
Face recognition is one f the main function of the proposed smart security system. Unlike
the previous system , where the detection of the human presence in the room is only done by the
sensors, the new system will start to detect the face of the intruders when the sensor detect the
human presence. In order to detect the face, many already developed solutions can be applied.
OpenCV library can be used for this function in the system. Certain addons for the face
recognition can be added to the library. The user interfaces of the proposed system will ge
developed using JavaScript. In this case, clmtrackr can be used to for model coordination of the
face during the face recognition process.

5SMART SECURITY
Figure 2: Face coordination using clmtrackr
(Source: Artem & Vasyl,2017)
The identification of the human face is very complex, moreover, sometimes human faces
seems similar. In this case, the face recognition system in the smart security system can be
divided into three sections-
 Elaboration of the intensity of the image.
 Processing the video streaming.
 Collecting additional data image.
Automatic decision making of the process:
The automated decision making of the proposed system is based on the application of
neural network. Face recognition system detects the face by coordination of the faces and
developing the geometric structure of the face. In case, if the face pattern is not match with the
face pattern familiar with the system , the next step is decided by the system itself. The decision
making of the system based on the data given by the face detection system along with the past

6SMART SECURITY
data stored in the system, The machine learning implemented in the system will calculate the
presence of the residents in terms of hours of the day and date. The calculation of the probability
of presence of the residents in the hose is determined by the neural network.
Training of the neural network:
The training of the neural network is done by collecting the data in the normal day long
with the presence of the house owners (Zhang, Shan & Huang,2015). Any disruption in the
received data when compared with these collected data can be regarded as the abnormal situation
for the neural network.
Type of neural network used: Neural network with controlled learning is used. NN
technologies can be used in this case.
Size of neural network: Nine input neurons and two output neurons are used in this
case. Input neurons are dedicated to the communication channels. Two output neuron are used by
the security system software.
Figure 3: Neural Network Structure

7SMART SECURITY
(Source: Artem & Vasyl,2017)
Usefulness of the system:
The system has proved its usefulness as it also maintains a sound data integration by
providing the message alert through email and messages. The children and the old age people
can feel safe alone in the house as the system sends the updated security information about the
house to the other residents of the house.
Future modification of the system:
The smart security system has proved its effectiveness in maintaining the security of the
house. However, there are certain issues those can be highlighted regarding the system. The
implementation cost is high, which can be reduced so that everyone can implement the system in
their house (Mosenia & Jha,2017). In the technical aspect the modification can be made in
decision making field. The modified system can include the users of the system directly and the
decision can be made by the machine considering the suggestions from the house owners. The
possible flaws in the automation system is needed to be found out, as the intruders can use the
advanced technology to exploit the flaws of the system.
Evaluation of the system, weather it is successful or not:
Smart security system is not a new concept. However, the main objective of this
implementation is to develop a system which will be intelligent enough to detect the intruders
along with that the system will inform the residents of the hose about the trespassing that is
happening in the house. The system will be helpful for the resident as well as it will be beneficial
for maintain the laws. However, the success of the system depends on the implementation of the
system .It is assumed the proposed security system will be successful as it is advanced in
technical aspects and eliminate the drawbacks of the previous security systems.

8SMART SECURITY
References
Artem, K., & Vasyl, T. (2017, July). Structure and model of the smart house security system
using machine learning methods. In Advanced Information and Communication
Technologies (AICT), 2017 2nd International Conference on (pp. 105-108). IEEE.
Ateniese, G., Mancini, L. V., Spognardi, A., Villani, A., Vitali, D., & Felici, G. (2015). Hacking
smart machines with smarter ones: How to extract meaningful data from machine
learning classifiers. International Journal of Security and Networks, 10(3), 137-150.
Bakar, U. A. B. U. A., Ghayvat, H., Hasanm, S. F., & Mukhopadhyay, S. C. (2016). Activity and
anomaly detection in smart home: A survey. In Next Generation Sensors and
Systems (pp. 191-220). Springer, Cham.
Jacobsson, A., Boldt, M., & Carlsson, B. (2016). A risk analysis of a smart home automation
system. Future Generation Computer Systems, 56, 719-733.
Mosenia, A., & Jha, N. K. (2017). A comprehensive study of security of internet-of-things. IEEE
Transactions on Emerging Topics in Computing, 5(4), 586-602.
Shen, V. R., Yang, C. Y., & Chen, C. H. (2015). A smart home management system with
hierarchical behavior suggestion and recovery mechanism. Computer Standards &
Interfaces, 41, 98-111.
Simpson, A. K., Roesner, F., & Kohno, T. (2017, March). Securing vulnerable home IoT devices
with an in-hub security manager. In Pervasive Computing and Communications
Workshops (PerCom Workshops), 2017 IEEE International Conference on (pp. 551-556).
IEEE.
Zhang, J., Shan, Y., & Huang, K. (2015). ISEE Smart Home (ISH): Smart video analysis for
home security. Neurocomputing, 149, 752-766.