Chili plants are one of the horticultural crops with a fairly high selling value, but they require special attention in the cultivation process. The growth of chili plants is greatly influenced by the availability of water, both lack and excess water can interfere with the fertilization process and increase susceptibility to pest attacks (Imtiyaz et al., 2017). To support optimal growth, chili needs a growing medium with an appropriate level of humidity. Improper watering, both in terms of quantity and time, can negatively impact plant productivity. Therefore, the timeliness of watering is an important factor that

needs to be considered. Even if done regularly, watering that ignores the elements of time, temperature, and humidity can reduce crop yields.

The ideal watering time for chili plants is in the morning at around 07.00 and in the afternoon at 17.00 WIB (Mootalu et al., 2022). Based on the age of the plant, water needs vary, namely at the age of one month around 0.111 liters per day, at the age of two months 1,148 liters per day, at the age of three months 1,223 liters per day, and at the age of four months it remains around 1,323 liters per day. In addition, chili peppers are also susceptible to pests such as caterpillars and aphids, which can cause serious damage to crops

Chili plants require special attention in cultivation, this special attention is paid so that chili produces abundant and quality fruits, this plant is a plant that is often targeted by various pests that can damage growth and reduce crop yields

The Internet of Things (IoT) is a technology that allows communication between machines through internet connections often referred to as M2M (Machine-to-Machine) with humans playing the role of users as well as managers. IoT consists of three main elements, namely physical devices equipped with IT modules, internet connectivity, and cloud-based data centers to store information. This technology continues to enable continuous connections between devices, machines, tools, and other physical objects through network sensors and actuators. In its operation, IoT devices connected

to the internet collect data stored as Big Data to be processed and analyzed as needed. With this capability, the device can monitor its own performance and independently take actions or decisions based on the information obtained. Collaboration between integrated machines creates an intelligent, efficient, and automated ecosystem, so that it can optimally improve various aspects of human life.

The initial idea of IoT was first introduced by Kevin Ashton in 1999 in one of his presentations. Since then, many major companies in the tech world, such as Intel, Microsoft, Oracle, and others, have begun to pay great attention to IoT development. Technologists predict that the influence of IoT will be the next big revolution in the world of information technology. This optimism arises because IoT holds various extraordinary potentials that can continue to be explored and utilized to create innovative solutions in various areas of life. There are several stages of how IoT works, namely sensors, data processing, connectivity, and actions, as seen in figure 2.10 (Sawitri, 2023).

In June 1965, the concept of fuzzy logic was first introduced by Professor Lotfi A. Zadeh of the University of California (Kaswidjanti, Aribowo, & Wicaksono, 2014). The basis of fuzzy logic is fuzzy set theory. According to Edy Victor Haryanto (2015), this theory states that the degree of membership of an element in a set is not only limited to the values of 0 and 1 (not members or members), but can be between these ranges. Thus, there is a logical "gray area" between the values of 0 and 1.

The ESP32-CAM is an ESP32-based microcontroller module equipped with an integrated camera, making it a very popular choice for applications that require the ability to shoot or video in Internet of Things (IoT) projects. Especially in this Final Project, ESP32-cam is used as a monitoring of the condition of chili plants and the surrounding environment.

DS18B20 sensors are analog sensors designed to measure temperature with high accuracy and simple data output via the 1-Wire analog communication protocol. These sensors are widely used in various applications such as temperature control, refrigeration systems, environmental monitoring, and IoT projects.

Soil Moisture Sensor is an electronic device used to detect soil moisture levels. This tool works by measuring the electrical conductivity of the soil, where the level of soil moisture affects the ability of the soil to conduct electricity. The

more water content in the soil, the lower the electrical resistance, so the output value of the sensor becomes higher. Soil Moisture consists of components with two metal rods that are embedded into the soil. These two metal rods are used for probes and electrodes to measure the electrical resistance of the ground. This soil moisture sensor has the following specifications:

The value range of ADC is 1024 bits ranging from 0 – 1023 bits.

Figure 6. Soil Moisture Sensor Module

The YF-S401 flow meter sensor is a Hall Effect-based flow sensor module commonly used to measure the rate of liquids, specifically water, in closed pipes. The specifications of this sensor are as follows:

Operating Voltage : 3.3 V – 5 V Working Current : ≤ 15 mA Measurement range : 1 – 30 L/min Accuracy : ± 10

Signal Output: PWM Diameter Pipa : ½ inch (G1/2)

Maximum pressure : < 1.75 MPa

Codular is a drag-and-drop based Android app development platform that allows users, even without a programming background, to create Android apps easily and quickly. Using the visual interface, users can add elements such as buttons, images, and text inputs and structure the app's logic using visual code blocks, which resemble puzzle pieces that can be combined to form a program flow. Kodular provides a variety of components, ranging from user interfaces, sensors, to internet-based services, and supports integration with

third-party services such as Firebase, Google Maps, and advertising. The platform allows developers to test apps directly on Android devices through live testing features, as well as offering app monetization capabilities through ads and in-app purchases.

MQTT (Message Queuing Telemetry Transport) is a lightweight communication protocol based on a publish-subscribe architecture designed specifically for machine-to-machine (M2M) communication and Internet of Things (IoT) applications. In its mechanism, MQTT sends data in the form of a message to a topic, which is a kind of category or virtual address used to group and convey data. The publisher sends a message to a specific topic, and the broker forwards that message to all subscribers who have subscribed to the same

topic. In this way, communication between devices becomes more flexible, efficient, and not directly dependent on each other. For example, a temperature

sensor can send data to a sensor/temperature topic, and the monitoring device only needs to subscribe to that topic to receive real-time temperature data (Pratama et al, 2023).

The measurement calculation in this study was carried out using the relative error method. This method is used to find out how close to the actual value obtained from a standard measuring device. In other words, relative error indicates the level of accuracy of the sensor in reading a physical quantity when

compared to a reference measuring instrument. In the process of measuring a physical quantity, be it temperature, humidity, water discharge, or other parameters, it is important to evaluate how accurate the sensor or measurement instrument used is. One of the commonly used methods is the calculation of relative error.

Relative error provides information about the extent to which the sensor reading results are close to the actual value obtained from a reference or standard measuring instrument. Thus, this method not only shows the difference in values, but also emphasizes the proportion of errors to the reference values. This is important because the same error can be considered small at large measurement values, but significant at small measurement values. (Bakshi & Bakshi, 2009).

(1)

Then to find the average value of the error from the entire data to be tested is as follows:

(2)

In the fuzzification system focused on finding the membership function, it has two inputs , namely temperature and humidity. Each of these inputs comes from the input of each sensor, such as the DS18b20 sensor to measure soil temperature and the soil moisture sensor to measure soil moisture. On the temperature variable, marks (D) are given for cold, (N) for normal, (P) for heat and (SP) for very hot. The temperature input variables are shown in the table as follows:

Table 1. Temperature Input

The following is an equation from the manual function looking for the parameters of fuzzyification:

From the table above, it can be concluded that for cold temperatures it is in the value range of 15-22 °C, for normal temperatures it is at 22-28 °C, for hot temperatures it is in the value range of 28-35 °C and for very hot values it is above ≥35 °C.

The soil moisture variables were taken based on the soil moisture sensor parameters by initiating the variables very dry (SK), dry (K), Normal (N), Wet (B). The following is a table of soil moisture input variables:

Table 2. Moisture Input

The following is an equation from the manual function looking for the parameters of fuzzyification :

The data that has been stored in the broker's MQTT is then connected to Kodular to be displayed in the form of real numbers. The creation of the application interface is carried out using the drag-and-drop method on the Koular platform, so that the development process becomes more efficient and easy to manage. Through this application display, users can monitor various important information in real-time, including:

In addition to the monitoring feature, the application is also equipped with additional menus such as live camera streaming for visual observation, as well as a plant information database that includes plant life, plant type, and planting season. With this integration, users can monitor the condition and development of crops thoroughly in one platform.

This table presents data on the test results of Fuzzy Logic-based automatic plant watering systems and the Internet of Things (IoT). Each row shows measurements based on the date and time of data collection, as well as input values from the sensor, the output of the watering results, and the difference or error between the simulation results in Matlab and the microcontroller.

Table 4. Fuzzy Logic Testing Data

The following is an average calculation of errors from the temperature data above.

For the example of rules with a temperature of 25 and a humidity of 78%, the results of the fuzzy mamdani calculation of 78.7 mL were obtained as the output of the amount of watering. To see the dimensions on the matlab see the image below.