Reducing mobile data usage and data volumes sent to the cloud

As the devices around us get smarter, the amount of data sent to servers over mobile networks keeps increasing. According to various sources, a single highly automated autonomous vehicle can generate up to 5 terabytes (TB) of data per hour, and in the future, more and more of this data will be sent to the cloud. Add in all the different IoT devices not connected to a WiFi network, and the amount of data sent over mobile networks will increase even more.

Why should we aim to reduce the amount of data we send over mobile networks?

Reducing the amount of data we send to the cloud can save both money and the environment.

According to cable.co.uk, the average cost of 1GB of mobile data in 2022 was $2.06 in Europe and even higher in North America at $4.98. This may not seem like a lot to the average phone user, but it adds up quickly for a company with hundreds or even thousands of devices regularly sending large amounts of data to the cloud. There is also the cost of storing the data, making it even more important to consider how much data is uploaded and stored.  

Data storage also has a large carbon footprint. Keeping the data storage systems up and running requires a lot of energy – approximately 40% of the electricity used in data centers is for cooling. While some data centers plan to become carbon-neutral using renewable energy, not all have the resources to do so.

We have reached a point where a single data center can use as much electricity as 50,000 homes. Storing only the data you really need in the cloud would help reduce the need for even more data centers in the future.

How can we reduce mobile data usage and the amount of data stored?

Data Compression

Data compression is one of the most common ways of reducing data size while retaining all the necessary information. It involves re-encoding information using fewer bits than the original data. There are two main types of data compression – lossy and lossless. The main difference between the two is that lossless removes statistical redundancies, which means no information is removed, while lossy compression removes unnecessary information and reduces the complexity of existing information. 

Therefore, lossy compression allows for much higher compression ratios at the cost of possibly lower data quality. Lossless compression formats include ZIP, RAR, 7z, FLAC, and PNG; some lossy compression formats include JPEG and MP3.

Data compression is most commonly used in audio and video processing, with specific examples including security cameras in remote locations, surveillance drones, and autonomous vehicles. In all of these cases, data compression helps reduce the large amounts of video feed data that need to be sent to the server, offering significant savings in both data costs and server space.

Send only relevant data to the server

IoT devices collect a lot of data necessary for the system to work, but not all of it needs to be sent to the server and stored there. Take an accelerometer, for example. Many IoT devices use accelerometers to detect motion but don't actually need the readings for anything. Therefore, collecting and sending the accelerometer readings to a server is unnecessary.

Another way to use this method is to choose the appropriate frequency for collecting and sending the data. Temperature monitoring is an important part of many IoT systems, either to protect the system from failure or to have an overview of the conditions of the space around it. 

However, significant temperature changes don't occur in seconds or even minutes, so sending temperature data to the server every half hour or hour is sufficient. In other cases, it may be enough to send data and an alarm only when a set limit is reached, saving even more.

This is the most common method used with all kinds of sensor devices – environment monitoring devices, weather stations, and motion detection devices.

Send processed data to the server

As smart devices become more powerful, processing data at the device level becomes easier than sending large amounts of data to the cloud. In this case, processing can mainly be filtering, performing calculations, or using a machine learning or artificial intelligence algorithm. The goal of processing is to reduce the data and transform it into a more understandable form before sending it to the cloud. 

Today, most sensor IC manufacturers already offer ICs capable of processing different algorithms, making it even easier to add device-level data processing to your system. Some of these ICs are widely used in smartphones and fitness trackers to count steps or recognize other movements at the device level.

Autonomous vehicles also use embedded computers like NVIDIA Jetson to analyze the data on board using AI to detect situations and make decisions accordingly. In some of these cases, doing this on board not only saves data but also time by not having to send time-critical data to the server and then send the decision back.

Send data to the server over WiFi

Finally, perhaps the simplest solution to high data usage is to send data to the server over WiFi instead of using cellular data. In many systems, there are times when the device is in WiFi range, even if it is out of range most of the time. 

In cases where the data is not time-critical, it makes sense to store the data on the device until there is a WiFi network and then send all the data to the server, rather than constantly doing it over cellular networks. You could also use a combination of cellular data and WiFi by setting conditions on the system so that the time-critical data is sent to the server immediately, and the rest is sent when a WiFi network becomes available.

This method is often used for system maintenance. An alarm or error code is sent to the server when something is wrong with the system, but the specific data and logs about the problem can be sent to the server when the device is taken in for maintenance, where it can be done over WiFi.

In a world where enormous amounts of data are generated every second, it is easy to forget that not everything needs to be sent to the server and stored there just in case it’s needed in the future. Thinking about what the data is used for and finding the best way to reduce it is both cost-effective and good for our environment.