The success of the IoT solution is very dependent on how the data consumption per each endpoint would be precisely considered. From one side, the IoT solution provider must utilize the significant amount of data for further analysis and decision making, and from the other hand the elements such as battery life of the device, the capacity of CPU, the costs of transferring and storing the data must be carefully analyses.
In this short post, we are going to highlight the primary decision criteria for the allocation of the right data volume for each device involved in connectivity.
Depending on the industry and business sector you act in, the following factors affect the data consumption and transfer:
- The type of device – Smart meters, Industrial equipment, Sensors, etc.
- Sample rate - frequency of samples per second
- Resolution – the data size in each sample
- The signal conditioning – defines the (ir)regularity of initiation the data transmission based on different factors (time of the day, season, etc.)
- The amount of local data storage – the size of the hard drive, keeping the processes, etc.
- The power consumption requirements of the device - the battery life, accessibility of the device, etc.
So, let’s make a simple example of the smart sensors that are responsible for notifying the quality elements of water tanks in the rural areas with the deficit of drinking water:
- Device – Water sensor that measures Temperature, pH, etc.
- Sample rate – 10 kHz (2x of range 0 – 5 Hz)
- Resolution – 16 bit (2 bytes)
- The signal conditioning – each one hour
- Amount of data: 20 kHz x 2 bytes = 40 kb/sec
- The # of sensors 1000
- Amount of data: (40 kB/sec x 1000 sensors) x 1 = 40 MB/hour
The rough calculation shows that each sensor must have in readiness at least 28,8 MB data volume. In addition to this amount, we must also consider a different type of additional small data requests initiated both by endpoint and the central system. The final approximate data volume for each device can be 30 MB.
Sometimes, especially at the early adoption stage of business, the usage pattern of the devices generating the data might not be equally distributed, and instead be followed with the different fluctuations. In this case, it would be expedient to start with the Pay as You Go pricing plan and monitor the data generation and transfer regularity.