The Internet of Things

Smart homes are becoming more popular today because of two reasons. First, the sensor and actuation technologies along with wireless sensor networks have significantly matured. Second, people today trust technology to address their concerns about their quality of life and security of their homes (see Figure 8).

Figure 8 

Block diagram of a smart home system.

In smart homes, various sensors are deployed, which provide intelligent and automated services to the user. They help in automating daily tasks and help in maintaining a routine for individuals who tend to be forgetful. They help in energy conservation by turning off lights and electronic gadgets automatically. We typically use motion sensors for this purpose. Motion sensors can be additionally used for security also.

For example, the project, MavHome, provides an intelligent agent, which uses various prediction algorithms for doing automated tasks in response to user triggered events and adapts itself to the routines of the inhabitants. Prediction algorithms are used to predict the sequence of events in a home. A sequence matching algorithm maintains sequences of events in a queue and also stores their frequency. Then a prediction is made using the match length and frequency. Other algorithms used by similar applications use compression based prediction and Markov models.

Energy conservation in smart homes is typically achieved through sensors and context awareness. The sensors collect data from the environment (light, temperature, humidity, gas, and fire events). This data from heterogeneous sensors is fed to a context aggregator, which forwards the collected data to the context aware service engine. This engine selects services based on the context. For example, an application can automatically turn on the AC when the humidity rises. Or, when there is a gas leak, it can turn all the lights off.

Smart home applications are really beneficial for the elderly and differently abled. Their health is monitored and relatives are informed immediately in case of emergencies. Floors are equipped with pressure sensors, which track the movement of an individual across the smart home and also help in detecting if a person has fallen down. In smart homes, CCTV cameras can be used to record events of interest. These can then be used for feature extraction to find out what is going on.

In specific, fall detection applications in smart environments are useful for detecting if elderly people have fallen down. Yu et al. use computer vision based techniques for analyzing postures of the human body. Sixsmith et al. used low cost infrared sensor array technology, which can provide information such as the location, size, and velocity of a target object. It detects dynamics of a fall by analyzing the motion patterns and also detects inactivity and compares it with activity in the past. Neural networks are employed and sample data is provided to the system for various types of falls. Many smartphone based applications are also available, which detect a fall on the basis of readings from the accelerometer and gyroscope data.

There are many challenges and issues with regard to smart home applications. The most important is security and privacy since all the data about the events taking place in the home is being recorded. If the security and trustworthiness of the system are not guaranteed, an intruder may attack the system and may make the system behave maliciously. Smart home systems are supposed to notify the owners in case they detect such abnormalities. This is possible using AI and machine learning algorithms, and researchers have already started working in this direction. Reliability is also an issue since there is no system administrator to monitor the system.