A recent study of Cisco shows that the annual global cloud trafﬁc will reach up to 14 zettabytes yearly by 2020 . Although energy costs constantly decrease while performance of electronics rises as well as its efﬁciency, the transfer, processing and storing of data will always cost a certain amount and because of the huge amount of data, a small decrease in the volume of data could save a lot of the costs to its users.
Currently in most cases 100% of sensor data is sent directly to the cloud without any ﬁltering and the data is processed there. However often a large percentage of the data is very noisy, contains invalid values or contains no valuable information to the user. With edge computing, the raw data is processed already on the IoT device, therefore close to the source, and data or gained information is only sent to the cloud after being processed and cleaned. Although the percentage may vary on the usage of the IoT device, it is estimated that about 30-50% of the data transfer and storage costs can be saved with edge processing.
It must be noted that internet of things devices often have relatively low processing capacity and in case too much information must be processed by it,it may create an overhead that has to be either handled by sending most of the data to the cloud and processing it there, or simply ignoring the overhead data. Both cases would have their drawbacks, therefore the frequency of data collection by the device from the sensors must be adjusted accordingly.