Graph Technology in IoT: For Improved Data Management & Customer Service
# What is IoT?
There must be times you’re suddenly reminded that you forgot to turn off your Hair Iron when leaving the house out of nowhere. (Yes, your TV, Humidifier, and anything electronic.) And it’s always this grocery shopping myth that you always can’t remember what was left in the fridge when you want to drop by a store on your way home.
How convenient would it be if you could control your home devices from outside your house and manage your grocery list from the fridge automatically?
The technology that makes these possible is the IoT (Internet of Things.)
IoT is a technology that enables everyday ‘things’ to communicate with the internet. Through this technology, various home appliances and devices like Smartphones can share information with each other and send the data generated in communication. With the IoT technology, it might not be wrong to say that we can now communicate with our devices.
*Things: Devices, sensors, machines, communication between people, the internet, etc.
IoT– by connecting devices and appliances we use, allows us manage them with an integrated system to help make our daily lives much more convenient. The IoT technology is rapidly increasing its variety and are widely and deeply influencing our lives every day.
# Graph Database– new technology in IoT implementation
Graph databases, consisting of graph connection, are a database that is most suited to storing and managing data. The data collected from the IoT devices are data made from connections between devices and sensors and are represented in con
Graph Database is a solution that implements technology to prepare the best data operating environment, such as storage, operation, and processes while maintaining the original form of such graph data in a real-time processing environment.
Therefore, when building the IoT data system in which the data form consists of connections, using the graph DB will maximize performance and efficiency. Here are the two ways to utilize the Graph DB;
- Utilizing Graph Technology from the data management perspective
- Utilizing Graph Technology from the customer service perspective
I will explain in detail the usage point of both aspects that the combination of IoT and graph databases can provide.
# Utilizing Graph Technology from the data management perspective
As the IoT technology develops, the number of IoT devices and users increases, and the data generated by the IoT system also increases accordingly. Under these circumstances, efficient IoT data management is essential in cost reduction for companies that utilize IoT technology.
In order to manage the connected IoT data in the connected form well, below are two aspects to consider.
1) Configuration management of IoT data generated by being connected
In IoT devices and applications, a large number of data transmissions between sensors are connected and generated. Again, Graph Database is the most suitable for processing and managing these large transmissions.
When managing the data generated by many objects connected to a relational database, the number of joins in the tables of each object increases. Similarly, in IoT data, as the number of connected IoT devices increases, the amount of computation increases and becomes overloaded. In IoT data management, this generates significant costs.
On the other hand, a graph database is storage that adds a newly entered object as a node and stores the relationship with another object /node in connection with this by connecting it with a line(edge). We can say that this is a database that can intuitively store data in the form of connections generated by the IoT system.
Therefore, if you manage IoT data with a graph database, you will be able to utilize the meta-information of IoT more efficiently by graphing it and proceeding with a quick search when searching for data in the form of connections.
2) Processing various types of data generated by IoT devices
The number of IoT device types and applications is increasing every day, and each device generates various data types. With the increasing number of devices, the types of data the new devices will produce are unpredictable. Therefore, data storage with flexible data management is required.
If you use NoSQL databases, you may easily accommodate the data generated from many devices that are difficult to predict. NoSQL databases have a schemaless structure, and it does not require defining all keys and data types in the data model in advance. Therefore, you can store the data generated by many types of IoT devices the way they are input.
In a graph database, a representative NoSQL database, the schema does not exist and has excellent usability to manage IoT data. It allows for managing various data types flexibly with no additional work by engineers.
#Utilizing Graph technology from the customer service perspective
For companies using IoT technology, it is important to manage data efficiently, yet it is paramount to consider the quality of service provided to customers using IoT. In order to improve the quality of service, accuracy, personal customization, convenience of use, and more must be considered. Using graph databases on IoT technology will undeniably help provide customers a quality service.
Seeing from the customers’ point of view, here are the three factors improved through the services provided with graph technologies.
- IoT system failure monitoring service(accuracy)
IoT data consists of connections between devices and sensors, and failures can occur in many devices and sensors, making receptions impossible. Finding the cause of these failures in a responsive manner is a crucial factor in IoT service quality.
When failures occur, the relational database(RDB) must perform joins of devices and sensor tables and find out which devices are causing problems. Depending on the number of devices and sensors, a large calculations may occur, taking an immesuarable time to find the cause of the failure. On the other hand, if you manage data with a graph database, you would only need to perform the search of the devices and sensors connected to the error device leading to a little to no computation required.
Similarly, the process to be pre-taken to quickly find out the problems that occur in the IoT system by managing the IoT data with a graph database is to model the IoT data in the form of a graph.
There are various types of data that may be stored in the IoT system, and of the many data types, most can be categorized into the seven types.
<Types of data that can occur in IoT>
- Characteristics and status of the device
- Data which goes back and forth in the device
- Application program which can access the device(application)
- Sensor
- Manufacturing Company
- Device Department
- Device User
- Command Voice Data, etc.
The defined 7 types of data can then be connected with each other using edges and structured into a graph model with the schema as below.
Through intuitive graph modeling, the failure events are detected quicker and improve the quality of service;
- An intuitive search of the department in charge connected with the device
When there is a problem with a specific device, you can search the application connected to that device and the department in charge of that all at once, hence, improving responsiveness in service. - Efficient check of device information
Takin battery devices, for instance, understanding the efficiency and status is an essential factor to running the IoT service smoothly. Therefore, by searching for the device node connected with the node whose device status is discha
2. Recommendation of customized IoT devices through prediction of user behavior patterns
Users that use the IoT devices connect the devices based on their personal preferences and make their daily lives more convenient. The way to improve the quality of service for these users is to find out about the potential needs through identifying the user behavior patterns. Will identifying the devices and the functions needed for their daily lives in advance improve the service quality? By identifying the user patterns, the recommended devices and functions customized for particular individuals are searchable, improving the service experience far better.
Today, it is crucial for data administrators and companies to be able to predict user behavior patterns.
To identify the behavior patterns of IoT users, you need to build a behavior-based graph first. The behavior of user A using the IoT device is expressed as follows.
User A has a pattern of turning on the lights and opening the blinds every morning at 7:00 AM as the alarm hits. In order to recommend the IoT device which would be good to use in the morning for the user A, we need to find the user pattern with a similar behavior pattern as the user A. Through expanding the behavior patterns of the user with the ‘alarm clock -> lights’ in the morning, found the below.
As shown, this user follows the ‘Alarm Clock-> Light’ pattern, with added speakers and blinds actions. Learning from this pattern, User A might also like to purchase speakers. If user A already has a speaker, then the IoT device connection service to auto-play the music in the morning is highly suggested to improve user A’s satisfaction.
3. IoT system automation using a knowledge graph
The IoT users usually operate the devices by giving direct voice commands or pre-setting the applications in advance. However, in order to improve the level of convenience of use of the IoT service, it is necessary to make sure that the devices operate automatically with simple commands without direct device commands and pre-settings by users.
This can also be performed by building users’ command data collected from the IoT system into a knowledge graph, producing automated IoT services. Based on the knowledge graph built, you can also combine it with machine learning to freely analyze the life cycle data and understand the IoT users. IoT data-based knowledge graphs learn the IoT system behavior to make the interactions between humans and devices more natural and intuitive.
Let’s look into the automatic temperature control in a building as an example. Let’s say that the data obtained from the building’s temperature sensors include the temperature value and the sensor location in the building. You can identify the change in the temperature value recorded on the temperature sensors and learn what factors change the indoor temperature settings using a knowledge graph.
When someone in the building says, “It’s too hot here!” the words “here” and “hot” can be extracted. By approaching the node “here” in the built knowledge graph, you can find the sensor closest to the person’s location.
It searches for connected nodes to check if the temperature value measured with a found sensor is high, and commands to lower the temperature.
When you find out that the user is located at the window sensitive to the actual temperature, you can automatically adjust the blinds or turn on the air conditioner to lower the temperature.
As shown above, building human instruction languages generated through IoT devices into a knowledge graph can enable IoT automation. IoT services can then be more convenient to users. Furthermore, the approach can be expanded across industries such as retail and automotive.
#Conclusion
The methods to utilize and enhance IoT technology as a graph DB are as follows.
- Utilizing Graph Technology from the data management perspective
• Live configuration management of IoT data generated by being connected
• Processing various types of data generated by IoT devices - Utilizing Graph technology from the customer service perspective
• IoT system failure monitoring service
• Recommendation of customized IoT devices through prediction of user behavior patterns
• IoT system automation using a knowledge graph
Because IoT technology is closely linked to our lives, data security can also be considered to be an important factor. Recently, GNN, machine learning-based bottleneck phenomena, and anomaly detection are also researched using graph-IoT data. To use IoT technology more efficiently, the graph databases must be fully utilized.
