Blog

Introduction

The Internet of Things (IoT) refers to the communication and processing of data between numerous devices through an Internet network. IoT devices include sensors, digital actuators, and mobile terminals, excluding general desktops and laptops.

This architecture is regardless of the data quality that developers need to focus on, there are a lot more other things that need to be done to make the service itself work.

However, IoT data has slightly different characteristics.

Typically, large amounts of data occur sporadically in many places, making it very difficult to process with existing data solutions.

In this post, we will look at the types and characteristics of IoT data, and talk about the challenges that must be overcome to process IoT data.

The Internet of Things was once a term referring only to systems using RFID data, but with the development of ICT technology, more and more types of data have been included.

Then, there are various types of data, including RFID, and let’s look at their characteristics.

RFID is a tag that transmits and receives information recorded by radio waves and can be attached to or included in equipment. An RFID tag consists of an IC chip that stores data and an antenna that transmits and receives data, and the tag data is communicated wirelessly through a tag reader.

RFID is used in a wide variety of fields. For example:

• passport
• Cell Phone
• logistics management
• Inventory Management
• healthcare

Tags can be used in many fields as they become very inexpensive through mass production, but they are still more expensive than bar codes, so diffusion is slow in areas such as distribution.

When RFID is used in logistics, the movement trajectory according to the time series can be tracked based on the tag’s location information and time information.

Log data generated by numerous S/W and H/W plays a very important role in managing devices and software.

However, since log data is created in text format and can be automatically deleted when it reaches a certain amount, other methods are needed for long-term collection and analysis of data.

Since it is not structured data, a conversion process such as parsing of log messages is required to express it as a relational DBMS schema.

Log data is recorded in various formats depending on the program that creates it, so it is not easy to process.

As can be seen in the example of RFID, the location information of the place where the data is generated is very important for moving object data and meteorological environment data.

In general, location information is obtained using a global positioning system (GPS). GPS information obtained through several satellites can only know an approximate location due to its characteristics, and it is difficult to obtain accurate location information.

In special circumstances, more detailed information can be obtained using a local positioning system.

Positional data of non-moving equipment can also be treated as very important information.

For example, by combining environmental information such as temperature, humidity, and air pressure from a sensor floating on the sea with location information, it is possible to obtain information that is very useful for weather forecasts and disaster warnings.

Location and environmental data are being studied in convergence with technologies such as geographic information systems and mobile computing.

We live surrounded by numerous sensors. Mobile phones are also equipped with numerous sensors such as cameras, GPS, and accelerometers, and there are many sensors in factories and public sectors (roads, railways, ports, and airports).

Analyzing this sensor data can solve previously unactionable problems in various areas. Each sensor has a unique identifier, and records and transmits the read data value and measurement time together.

Data recorded in the form of <Timestamp, sensor identifier, sensor value> is stored sequentially according to the input time for later data analysis, and this is called time-series sensor data.

Not only sensor data collected from actuators that change in real time, but also control signal data for controlling the actuators are recorded in time series.

Since these data are data that is changing in real time, a large amount of data is generated, making it difficult to store and analyze. Afterwards, diagnosis can be made by analyzing existing data such as accident analysis, defect prediction, quality improvement, and production control.

When sensor data including time is collected, this data becomes historical data. The amount of data greatly increases according to the data collection cycle.

This is a problem to be solved with DBMS because the shorter the data collection cycle for detailed analysis, the larger the amount of data.