Iot-based Smart Irrigation Systems: An

Technology References

	169,177,182,187,193]
SIM900	[35,36,44,61,72,122,129,179,192]
NRF24L01	[124,131,173]
XBee S2	[119,121,144]
SX1276	[78,95,145]

_ESP82666 [34,43,47,56,64,69,71,82,84,86,87,94,97,111,113,114,116,119,123,126,127,129,158,160,166,168,

3. 
   Cloud Platforms
   

In this subsection, an overview of the most frequent cloud systems for IoT in smart irrigation solutions is going to be discussed. The two main storage systems used by the authors are traditional databases or cloud. In one article, the authors indicate that they store the data in a Raspberry Pi [124]. As it can be seen in Figure 26, a large percentage of the authors, more than 50%, have not defined the storage system they have used, there are 79 other papers where the storage system used is not specified.

Figure 26. Number of papers per storage system.

To be able to provide the new services that are currently demanded in IoT, it is necessary to use middleware. Through middleware, it is possible to connect programs that initially had not been developed to be connected to each other. Bandyopadhyay et al. [199] presented a classification of the IoT-middleware, based on the various features and interface protocol support. One of the reasons why middleware is needed to connect IoT devices, which are initially autonomous, is to provide cloud services. We can take advantage of existing middleware using it on different cloud platforms.

When we talk about the cloud, we identify a place where data, which is collected through sensors and transmitted to remote locations, will be stored and processed. In most of the articles, it is indicated that the data is processed in the cloud itself, and the end-users view all the information by connecting to the cloud. Cloud storage has been done on different platforms as it can be seen in Figure 27. In 24 of the reviewed articles, the authors specified that they store the monitored data in the cloud, but do not identify the utilized cloud platform (see Table 8). Most of the works in which the authors define the employed cloud platform, the utilized platform is Thingspeak (The MathWorks, Natick, MA, USA), with 14 papers. This platform is very intuitive and provides both free and paid options for storing, analyzing and displaying the data on different devices. Algorithms can be developed using MATLAB (The MathWorks, Natick, MA, USA) to generate alerts. However, there are proposals that use other cloud platforms such as FIWARE (FIWARE Foundation, Berlin, Germany) [75,149] and Dynamo DB (Amazon DynamoDB, Seattle, WA, USA) [117,159] with two papers each, and MongoDB (MongoDB
Inc., New York, NY, USA) [28], Ubidots (Ubidots, Doral, FL, USA) [5], Amazon (Amazon, Seattle, WA, USA) [91], M2X (AT&T, Dallas, TX, USA) [200], NETPIE (Nexpie Co., Ltd., Bangkok, Thailand) [43], SAP (SAP SE, Walldorf, Germany) [201], InteGra (Integra Network Services, Milford, MA, USA) [194], Firebase (Firebase, San Francisco, CA, USA) [154], InfluxDB (InfluxData, San Francisco, CA, USA) [176]. These less-used platforms are either more expensive, provide fewer services or are less intuitive than Thingspeak. Table 4 lists the cloud platforms that have been used in the reviewed articles.

Figure 27. Number of papers that employ each clouds service or platform.

Table 8. Cloud platforms.