Disclaimer: This is an example of a student written essay.
Click here for sample essays written by our professional writers.

Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UKEssays.com.

Motion Detection in Smart Home Environment

Paper Type: Free Essay Subject: Information Technology
Wordcount: 3324 words Published: 8th Feb 2020

Reference this

List of Tables

Table1: Sequence of state changes to recognise the activity 6

Table2: ANN input values for enter and exit of home 8

List of Figures

Fig1: Sensor values 6

Fig2: ANN network 7

List of Acronyms and Abbreviations

IoT  Internet of Things

HVAC Heating, ventilation, Air-conditioning

AAL Ambient Assisted Living

PIR Passive Infrared Sensor

ADL Activities of Daily Living

HMM Hidden Markov Models

ANN Artificial Neural Network

M2M Machine to Machine

www world-wide web or internet


Human motion detection is a technique which has a wide range of application both smart home and healthcare. Motion detection technology is far better than video surveillance technology which is a time-consuming reviewing process. By using motion detection, it saves both time and cost. In last few years it has gained lot of interests. Motion detection can be used for fall detection for elderly people, controlling HVAC (Heating, ventilation, Air-conditioning) system, lighting system, activity detection within home such as security purpose and AAL (Ambient Assisted Living). In this paper a motion system has been discussed based on currently available technology. In this approach PIR (Passive Infrared Sensor) has been used to identify the activity of the human. This technology is quite simple and non-instructive because the sensor nodes are connected to battery power and no additional work is required for insuring the power supply.


Number of physical devices rapidly increasing that are connected to the IoT (Internet of Things) applications which can improve our live style. It is estimated that by the year of 2020 connected devices to the internet will be 20 billion. Applications that are connected to the devices can be divided into three groups; 1. Industry domain 2. Smart cities domain and 3. Healthcare and wellbeing domain. Smart home which is a part of smart city are often mentioned in different types of survey of IoT. Remote control of HVAC which will provide home automation system is possible by connecting different types of devices to the internet that will be based on web or mobile application. Embedded processors and communication system with advancement of sensing technologies enhance the independent living services from health domain within smart homes. These services are known as Ambient Assistant Living (AAL). The main purpose of this system is wellness of elderly people, disable people and people with acute pathologies. Wellness of people can be observed by monitoring the Activities of Daily Living (ADL). This monitoring can be done using motion detection technologies. These activities could be sleeping, eating, walking, watching TV, toileting and entering & exiting home. Detected activities can be analysed to detect the patter in behaviour using machine learning techniques.


Motion detection is the process of detecting changes in various objects relative to its surroundings or changes in surroundings related to object. There are many types of motion detection such as Infrared (Passive and active sensors), Optics (video and camera systems), Radio Frequency Energy (radar, microwave and tomographic motion detection), Sound (microphones and acoustic sensors), Vibration (triboelectric, seismic, and inertia-switch sensors), Magnetism (magnetic sensors and magnetometers). For this project, infrared type of motion detection has chosen.

Get Help With Your Essay

If you need assistance with writing your essay, our professional essay writing service is here to help!

Essay Writing Service

Infrared motion detection technique is two types passive infrared type and area reflective type. In passive infrared type, the sensors are designed to cover wide area with the detection of human presence. The sensor, rather than emitting light such as from LEDs, detects the amount of change in infrared rays that occurs when a person (object), whose temperature is different from the surroundings, moves. The advantages of this techniques are it can cover wide area. On the other hand, the drawback of this techniques is that the sensing techniques cannot be set.

In area reflective type, sensor emit infrared light. Using the reflection of those rays, it detects the motion and distance of the object. The main advantage of this technique is that the distance can be set for sensing. On the other hand, the drawback of this technique is that it can be used in specific range.


The significance of this project is huge. There are many aspects where it has application.

• Building Automation

• Motion Detection

• Intrusion Detection

• Occupancy Detector

• Room Monitors

• Pet Detection

• Fire Detection

• Falling Person Detection


In this section research problem, research question and research aim have been discussed.


There are three major problem for this project; first of all, the activity detection has to be real time because sometime the sensor processor takes too much time for processing the activity. Secondly, the output must be efficient that means the motion detection have to be acute more than 99%. The final problem of this project is that the co-relation between the location of inside home environment.

1.3.2 Research aim

To overcome the first problem, activity detection methodology has to be set up which will provide real time processing value. To overcome the second problem, it can be worked upon different type of sensor. From these, the most efficient sensor has to be selected for the project. To overcome the co-relation of location inside home, the distance of each sensor has to be set by manually.


The research question is that how the motion will be done inside the home environment. And the second one is that what is the best approach for detecting motion? Why is that?

  1. Literature review

Now a day’s motion detection techniques have been widely researched around the world. Hong and Nugent focus on extracting segmentation data of each segment consecutive sensors events associated with complete activity. Among the lot of findings, smart home related topics have been presented in this section. Their detection includes using toilet, leaving house, going bed, taking shower and preparing meal. These have been done by considering the co-relation of location, object and sensor activity monitoring. They have proposed three type of approach for their project; location-based approach, model-based approach and dominant central model-based approach. All three approach have been showed similarly good performance.

Tao Gu has been presented how to avoid usual supervised learning phase in the machine learning process for activity recognition. This approach is on fingerprint based and testing this on various daily activity such as making coffee, watching TV, taking pill, washing cloths etc. The main idea is that retrieve the objects which has been used in specific activity and relevance their weight. Since the object may be used in shared activity so it is necessary to set a patter to contrast. Finally, their algorithm for contrasting patterns shows 91.4% precision.

Jie Wan has implemented a method to process sensor data in real time along with recognition of sensor data. Most of the researcher perform algorithm analysis offline using stored data set but in real time world data should be processed instantly so that if any action is required that can be taken instantly. The author has concentrated on data segmentation in real time using the correlation between time and sensor. Data segmentation based on daily activities in a smart home. Different types of algorithm have been tasted for recognition such as Bayesian network, decision trees, and Hidden Markov Models (HMM).

Reducing energy consumption is very important aspect specially when the device is on battery powered. Special attention must be given when the algorithm has been written for the machine or devices. Wang has presented a distributed event detection approach using self-learning threshold. Here, author has mentioned that energy consumption is one of the major challenges for these type of projects as off all the device will be battery powered. For his project, he has proposed timer-based node sleeping scheduling for the sensor.

  1. Research Methodology

Two approaches have been presented for this project. First one is sliding window and second one is artificial neural network (ANN).


This algorithm is based on the performance of two simple sensors which are PIR sensor which is a passive inferred sensor and hall effect sensor. PIR sensor is used for monitoring the presence and the hall effect sensor is used for monitoring the door is either open or close. Hall effect sensor is combination of two sensor, one is placed on the door and other one is placed on the door frame. Based on the magnetic field created by two part, the sensor produced different output. These sensors are connected to M2M devices.

For the PIR sensor if there is any motion detected by the sensor then it gives output as “1”, on the other hand if there is no motion detected then it gives output value as ‘0’. For the hall effect sensor if the door is open then it gives output value as ‘3’, on the other hand if the door is open then it gives output value as ‘0’.

To monitor the activity both sensor value is needed. Each sensor with two possible value in total four different combination of sensor reading.

Fig1: Sensor values

There could be six different activity.

  1. entrance when the door is closed beforehand, with closing the door afterwards,
  2. entrance when the door is closed beforehand, without closing the door afterwards,
  3. exit when the door is closed beforehand, with closing the door afterwards,
  4. exit when the door is closed beforehand, without closing the door afterwards,
  5. entrance/exit when the door is opened beforehand, with closing the door afterwards,
  6. entrance/exit when the door is opened beforehand, without closing the door afterwards.

To recognise the activity, sequence of state have to be pre-segmented manually.

Table1: Sequence of state changes to recognise the activity


Sequence of state changes






s1→s4→s3→s2→s1; s1→s4→s3→s1; s1→s4→s3→s4→s1







Coding for sliding door window,

1: procedure find activity

2: loop:

3: presenceValue ←value of presence sensor

4: doorValue ←value of hall effect sensor

5: State ←(presenceValue, doorValue)

6: if S tate! = lastStateInQueue then

7: Queue ←State

8: if queueS equence = activityPattern then

9: print Activity

10: goto loop.

3.2 Artificial Neural Network Based Algorithm:

ANN is a branch of machine learning based on replicated biological neural network which can be used to determine human activity. Nodes and connection between them are the key element of this algorithm. Based on learning sample, ANN adapts connection weight.

All the sensors relate to each other. They are also connected to M2M devices. One separate measurement is not enough for activity detection but also it needs to know the changes in sensor output in time. The layout of the network to specific form of a data which is being analysed need to be adjusted. The disadvantage of this algorithm is that limited number of stored samples.

Fig2: ANN network

Going through a training phase network learning is necessary. Pseudo random numbers are assigning in the learning phase. Input values used in this approach both for training and recognition phase are shown in Table 2,

Table2: ANN input values for enter and exit of home


Sensor State


the door is closed, no person is present in front of the sensor


the door is opened, no person is present in front of the sensor


the door is opened, person is present in front of the sensor


the door is closed, person is present in front of the sensor

ANN Coding:

1: procedure Find activity

2: Network ←create Neural Network

3: training(Network)

4: fileReading ←read presence and door values from file

5: loop:

6: Line ←one line of fileReading contains presence and door values with same timestamp

7: if Line! = null then

8: State ←stateSetter(presenceValue[Line], doorValue[Line])

9: Queue ←State

10: Activity ←activityRecognizer(Queue,Network)

11: if Activity! = no activities then

12: print Activity

13: goto loop.

First, neural network is created based on parameter then the training is executed. After that the loop start in which sensor data is brought into the ANN layer. Comparing between pre-segmented input value and output value, ANN takes the decision. For detecting the activity inside the home, it takes the values from the respected sensor and compare that with pre-stored data where the sensor is located.


Both algorithms have been tested offline, on static data set and online in real time. Precision and recall for sliding door window in offline testing are around 60%. On the other hand, ANN algorithm testing in offline precision and recall are              30% and 38% respectively. For ANN network precision and recall would be enhanced if larger data set is used in training. However, sliding door window is easy to implement and shows better performance. Even in online analysis, for sliding door window precision and recall are 76% and 86% respectively. In online analysis some of the data set is sent from M2M to gateway, whereas in offline testing the data set could not reach the destination. That’s why online testing shows better result. Finally, it is clear that that sliding door window best method for detecting motion inside home environment.

  1. Ethics:

Ethics are the customary norms and ways of behaving in a society. Whether an individual act ethically or unethically is the result of the complex interaction between a manager’s stage of moral development and several moderating variables including individuals’ characteristics, the organisations structural design, the organisations culture and the intensity of the ethical issue. People who have a strong morals sense are much less likely to do the wrong thing if they are constrained by rules, policies, job descriptions or strong cultural norms that disapprove of such behaviour. Conversely, very moral individuals can be corrupted by an organisational structure and culture that permits or encourages unethical practices. A researcher should maintain good ethical behaviour for the sake of this world and society. Because the principal concern of ethics is the human well-being. Morals are a reflection on those norms and the deliberate generation and adoption of principles which may well modify them. A researcher should also have to be moral, being moral conforming to the rules of right conduct. A moral society is a well-functioning society. Become a better, more perfect human being. Moral people are happy people.

  1. Conclusion:

Effective and convenient motion detection is presented in this work. Both methods detect the motion and give the output value based on pre-segmented data. The system is applicable both in home and office. By successfully implementing the project in home environment, remote control of HVAC, intrusion detection, fall detection, well being of elderly people and monitoring people with acute pathologies are possible. Currently, this project is unable to differentiate between the motion of a human being and a object. In future, there is a plan add this phenomenon, with this project.


  1. Al-Fuqaha, A., Guizani, M., Mohammadi, M., Aledhari, M., Ayyash, M.. Internet of things: A survey on enabling technologies, protocols, and applications. IEEE Communications Surveys Tutorials 2015;17(4):2347–2376. doi:10.1109/COMST.2015.2444095.
  2. Gartner, Gartner Says 6.4 Billion Connected ”Things” Will Be in Use in 2016, Up 30 Percent From 2015. Press Release; 2015. URL: http://www.gartner.com/newsroom/id/3165317.
  3. Suryadevara, N.K., Mukhopadhyay, S.C.. Smart Homes: Design, Implementation and Issues. Springer; 2015. ISBN 978-3-319-13556-4.
  4. Soulas, J., Lenca, P., Th´epaut, A. Unsupervised discovery of activities of daily living characterized by their periodicity and variability.
  5. Engineering Applications of Artificial Intelligence 2015;45:90 – 102. doi:http://dx.doi.org/10.1016/j.engappai.2015.06.006.
  6. Hong, X., Nugent, C.D.. Segmenting sensor data for activity monitoring in smart environments. Personal Ubiquitous Comput 2013; 17(3):545–559. doi:10.1007/s00779-012-0507-4.
  7. Wang, Y., Wang, D., Chen, F., Fang, W.. Efficient event detection using self-learning threshold for wireless sensor networks. Wirel Netw 2015;21(6):1783–1799. doi:10.1007/s11276-014-0885-9.


Cite This Work

To export a reference to this article please select a referencing stye below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Related Services

View all

DMCA / Removal Request

If you are the original writer of this essay and no longer wish to have your work published on UKEssays.com then please: