The Internet of Things

Implementing IoT

There are five areas that need to be addressed for a successful and beneficial IoT implementation. Let’s look at each of these with respect to capabilities and some practical application.

Sensors

The capabilities of sensors have increased by orders of magnitude in the past few years and, due to new technologies such as MEMS (Micro Electro-Mechanical Systems), costs have been reduced dramatically. Although there are several sensors for determining weight such as resistive load cells, vibrating wire, piezo electric, tuning fork and force restoration, today there are sensors for temperature, pressure, humidity, positioning, acceleration, direction, vibration, light intensity, level and fluid properties, just to name a few. Most of these sensors are easily integrated into systems using analog (or even digital) outputs. Accuracy is within a percent or two, if not greater, and mechanical mounting is simple with a clip, screw or even an adhesive mount. Connection to a high-tech programmable controller, such as the 1280 Enterprise™ Series allows the data to be collected and aggregated for decision-making.

Communication

The sensors involved in an IoT system need to communicate with the Internet in some way. This may involve wired, wireless or a combination of the two technologies. Since there are a number of technologies available, their individual strengths need to be considered.

The environment the sensor is exposed to needs to be taken into account. Temperature, humidity and exposure to outside conditions are factors. Can the communication be sent over wires from a fixed position? Does the sensor need to be mobile and still maintain connectivity? How critical is the timing of the information?

Security

This is an area that has been surprisingly overlooked in many implementations of an IoT solution. Even though there may only be a sensor transmitting information to a server over the Internet, there is still cause for concern. Any device connected to the Internet will be sending (and possibly receiving) data from one or more networks. This data is unencrypted, so it is readable by anyone that can get a connection. Rule number one for security on an IoT implementation (or any other connection to the internet) is: Never connect anything to the Internet without data security measures and encryption. This can be done relatively easily by using a VPN or firewall.

Data Aggregation

Once the data is being collected, then the decision has to be made regarding trends and how long to archive data. Much of this will be determined by the application and the goals set for that application. For instance, if the temperature of a tank is being monitored for out of tolerance high and low temperatures, then perhaps if the temperature stays within tolerance over the course of a month or two, the old data can be purged. If it goes out of tolerance a few times, the data should be kept to analyze the possible reasons why, and to correlate the temperature with other data to determine the cause. This is a big reason why IoT implementations fail (or succeed) to develop a return on investment (ROI): There is no preset goal or metric to collect the data in a way that allows the next section of IoT systems. Taking the time for setting goals and metrics before beginning an IoT solution will be time very well spent.

Decision making

You have the sensors in place, and the security in place. You are collecting data at a rapid rate and see some trends. What do you do about it? First, determine what decision you can make based on the data you have. The simple ones, such as temperature and humidity data can tell you if your process is taking those variables into effect (or if you need to). Does the temperature and humidity need to be controlled? Does the process need to be tweaked to account for these changes?

Now, think a step further—how about predictive maintenance? If the vibration of a motor is steadily increasing, at what point can/should the motor be rebuilt or replaced before it fails? Why is it drawing more power than usual? Should the filter in the heat exchanger be changed to allow the air to move more freely? Is the bearing in the motor failing? Is the load increasing for an unknown reason?

Here is where the true value of an IoT system becomes apparent. However, there are caveats. For instance, consider a warehouse that is mostly staffed with conveyors, robotic handlers and packaging machines for shipments. Only a handful of personnel are there to monitor the system and correct problems that may arise. A network of strategically placed sensors with infrared to detect people and control the lights/temperature are designed mostly for the benefit of the 15 human employees. The decisions made by the system are:

1. Keep the lights off, unless a person is in the area.
2. If a person is in the area, regulate the temperature to be between 65 and 80 degrees.

Sounds simple, right? Well, one day there are a few people sensed in the corner of the warehouse and the temperature is 93 degrees. The system turns on the lights in that corner and starts the air conditioning with the fan on low. Then, the system detects more people in the corner and the temperature is up to 104 degrees. The system turns on more lights and puts the air conditioner fan and cooling on high. There are now 35 people sensed in the corner and the temperature is now 128 degrees! The system now turns on more lights and puts the air conditioning on maximum.

Actually, there is a fire in the corner of the warehouse, no one is there, and we are fanning the flames with lots of cold, dense air! One of the first decisions in planning an Internet of Things, is to determine what decisions will be made autonomously and which decisions will require someone to get the information, verify whether something is abnormal and investigate. In this case, the fact that there were more people sensed than actual employees and the rapid rise of the temperature, despite the cooling system activation, should have activated an alarm.

In decision-making planning, all possible outcomes need to be addressed, even if it is simply “if this happens, then do that. However, if this is over a certain amount, then send an alert or shut down the process.”