Performance evaluation of communication methods on electric wheelchairs as assistive technology for persons with disabilities

Flex Sensor is a sensor that has a change in resistance as a result changes in the curve of the sensor. Generally, the principle of flex sensor is to take an action with the hand motion and to work based on the resistance value that transferred from the sensor into voltage via a board circuit. This sensor requires a voltage of + 5 V to work. The output resistance in this sensor can be given a voltage which can be read by the microcontroller. This sensor used to detect hand running movements in humans/other parts of the curve. The microcontroller converts data using an ADC (analog-to-digital converter), where the input data are obtained from the voltage that has been exposed to resistance. As robotic hand using flex sensor, Salman et al. (2020) were developed the hand motion with flex sensor to measure the angle of the finger by applying to five fingers. the voltage value of the flex sensor can be calculated by (1) Vout=Vin×RR0+R V_{out}} = {V_{in}} \times {R \over {{R_0} + R}} where V_out is voltage out of 2.5 V, V_in represents the voltage in of 5 V, and R_o+R_flat=R represent the resistance from the flex sensor. According to Saggio and Orengo (2018), the characterization of the flex sensor includes a single less than 100 μm and the flexible plastic is suitable for measuring of bending angle. The electrical value of flex sensor was affected by mechanical behavior (Saggio, 2012). A smart glove in this study was applied 5 flex sensors in the finger that can be seen in Figure 2.

Figure 2

In wheelchair system, Škraba et al. (2015) was developed the Arduino as microcontroller connected by USB to control the DC motor into digital output (DIG). Arduino in the electric wheelchair is for processing the converting the signal of the magnetic control with an 12C interface and installed on the circuit board. The advantage of this platform (Arduino) is each to use the digital and analog port as receive or sent the information data. This module is of great usefulness to control the motor movement of the wheelchair by voice controlled (Kian Hou et al., 2020). Arduino Nano has 8 pins as Analog inputs, labeled A0 through A7, each provide 10 bits resolution (i.e. 1024 different values). In general, Arduino can be applied in the IOT system to monitor agricultural filed (Yudhana and Kusuma, 2018; Yudhana et al., 2021a), medical field (Afsarimanesh et al., 2017; Yudhana et al., 2021b), chemical field (Bevc et al., 2017; Altintas et al., 2018), etc.

Arduino mega 2560 is a microcontroller board based on mega 2560 which has 54 pin digital I/O divided into 14 pins can be used 15 as PWM output, 16 Analog input, 4 UARTs (hardware serial port), 16 MHz crystal oscillator, USB connection, power jack, ICSP header, and reset button. This board also uses power connected to a computer with a USB cable or external power with an AC-DC adapter or battery.

nRF24L01 is a communication module long distance using the frequency band radio waves 2.4–2.5 GHz ISM (Industrial Scientific and Medical). nRF24L01 has speeds up to 2 Mbps with a choice of options date rates are 250 kbps, 1 Mbps, and 2 Mbps. In this system, nRF24L01 was applied to the data transmission and data reception as a communication module without wires between tools. According to Ganti et al. (2011), the advantage of nRF24L01 is easy and simple to control a few lines of code and easy to apply in predefined library. nRF24L01 also can be applied to wireless sensor to manage the robots using hand motion (Sivakumar et al., 2021). The performance of nRF24L01 was reviewed by Christ et al. (2011) and the use of multiple receivers can be reduced the amount lost packages consecutively. The transceiver consists of a frequency synthesizer integrated, power amplifier, crystal oscillator, demodulator, modulator, protocol engine, power output, frequency channel, and protocol setup easy to program via the SPI interface. The current consumption in the system is very low around 9.0 mA to 6 dBm and output power 12.3 mA in RX mode. Built-in Power Down and standby mode makes power saving with easy realization. The distance of the communication data is around 1 km (Prakash et al., 2020) with interface communication MCU I/O port. Yu et al. (2015) was reported that nRF24L01 has a high transmission power, high transfer rate, and available CRC error detection.

The principal work of DC motor depends on the current carrying conductor placed in the magnetic field (Gowthaman et al., 2021), which working like stator, rotor, commutator, and brushes. The varying of supply voltage, varying of the flux, the armature voltage and armature resistance are influence to DC motor. According to Sahu (2012), two ways to control CD motor using varying supply voltage and pulse width modulation technique, control DC using varying supply voltage requires analog circuit for digital system. While pulse width modulation is switching speed and pulse width or duty cycle.

Figure 3 shows the smart glove hardware in this study. The smart glove was installed a wireless communication system based on the nRF24L01 module as transmitter in which theses hardware are combined with the Arduino Nano module and flex sensor. This study was installed 5 flex sensors that can be described in Figure 8. All components are connected with a circuit/mini system and jumper cables. all components are connected with the ground port and voltage port both nRF24L01 and the flex sensor, then connected to the ground port and 3.3 V voltage port on the Arduino Nano board. Furthermore, the nRF24L01 component on the CSN and CE ports is connected to ports 7 and 8, then the MOSI, MISO, SCK ports are connected to ports 11, 12, 13 on the Arduino Nano board. The smart glove hardware design for this tool is divided into two parts, the first is the smart glove with the wiring, as shown in Figure 3.

Figure 3

Based on Figure 3 the electrical wiring in the smart glove created with five flex sensors and Arduino Nano. The flex sensor is applied to determine the resistance in the sensor curve (Trippolini et al., 2018). This study was developed from previous research (Yudhana et al., 2018) in which the flex sensor is to measure the level of curvature of the sensor and this sensor has an output resistance value proportionally with the level of flexibility. While the Arduino Nano is applied to control and process the system smart glove. In addition, both tools are used to control an electric wheelchair with data sent from the gloves to the chair using the module nRF24L01 (Yudhana et al., 2018). The electrical wiring in the smart glove can be described in Figure 4.

Figure 4

Figure 5 shows the electric wheelchair hardware is a means of assisting people with leg disabilities to be able to move from one place to another, either on a flat place or from a low place to a higher place (a place to climb). It is often also meant that electric wheelchairs are used to improve mobility for people with disabilities such as physically disabled people (especially those with leg disabilities), hospital patients who are not allowed to do a lot of physical activity, the elderly (elderly), and people who have a high risk of injury.

Figure 5

Figure 6 shows the wireless communication part of the electric wheelchair that makes use of the nRF24L01 module, with a maximum and minimum data transfer speed of 2 Mbps and 500 kbps, respectively (Bai et al., 2017). Furthermore, the platform used to process the electric wheelchair system is the Arduino Mega 2560, with the motor comprising of 500 rpm PG 45 and an encoder, named ems 30 H-bridge.

Figure 6

The software was designed by calculating and evaluating the error detection process and (CRC) algorithm, respectively. This flowchart is divided into two parts, namely the transmitter and receiver. During the transmission process set of check digits is calculated in each frame before it is added to the transmitter. In addition, the CRC calculation process is applied to analyze the receiver (Zhong and Hu, 2020). The CRC flowchart on the transmitter can be seen in Figure 7a. The framework developed in this research was implemented in the error detection method and the CRC algorithm. The flex sensors in smart glove input data are converted into binary form. Then Cyclic Redundancy Check (CRC) algorithm used for calculation between data from flex sensor and the amount of CRC value. Furthermore, after obtaining the CRC results, the value is converted into decimal form to minimize packet loss percentage during data transmission (Charanarur Panem et al., 2019). When the packet is sent in binary form, the data produced in binary digits (bits) increases the percentage of packet loss.

Figure 7

The CRC flowchart on the receiver part can be seen in Figure 7b. When the system receives a packet originated from the transmitter, a check analysis is conducted. This check ensures that the data received is not 0 before entering the next stage. Conversely, when the data received is 0, the system immediately ensures that it is classified as an error. The data are further converted into binary, after the system further analysis the data to ensure no error occurs. If the remainder is 0, it is identified as correct with assuming the reverse is the case. Finally, this system undergoes the CRC calculation process and generates a data quotient from the CRC calculation called the remainder.

The whole series of flowchart from the glove system as an electric wheelchair controller can be seen in Figure 7. These flowcharts are described the error detection and Cyclic Redundancy Cheek (CRC) calculation in electric wheelchairs automated using smart glove All instructions (Figure 8) are included onward (4), backward (O), left (L), right (V), and (5). when the system is ON, the first time the hand will bend its fingers according to the command codes that were previously set, then the system will match those curves whether they match the codes that have been set in the program. If not, the system does not respond and an experiment occurs. repeat on the curve of the fingers. If appropriate, the data coming from the system will be sent by the nRF24L01 module. If the electric wheelchair can respond, it means that the data from the system has been sent. If not, then back to the experiment of inputting command codes via finger.

Figure 8

Initially, this system hardware testing was carried out by checking the connection between the Arduino Uno and the flex sensor. The nRF24L01 module for transceiver and receiver was installed in electric wheelchair. The detail of electric wheelchair in this research can be seen in Figure 9. the testing of the packaging in this tool is carried out such as ensuring the screws, nuts, and bolts. These tools are checked to ensure all of the equipment installed with firm and unbeatable. Furthermore, electric wheelchair can be used with good equipment and easy to take the data collection. The people with leg disabilities can be directly seat and control wheelchair using smart glove in the right hand in 5 directions. A box control was added behind the electric wheelchair. All instructions in electric wheelchair are included onward (4), backward (O), left (L), right (V), and (5). when the system is ON. It is controlled by smart glove.

Figure 9

Error detection occurs in the process of transmitting data whose transmission process is not running properly. Error detection is an error detection process that is carried out when data is in the transmission process. The error referred to here is that there has been a change of 1 bit or more than one bit which should not have happened. The scenario of the error detection test is the flex sensor data from the transmitter in the form of an integer converted to a string. The value in the form of a string is converted from decimal to binary. Figure 10 shows the error detection test using the mathematical process with the implementation of cyclic redundancy check (CRC) algorithm. A binary number is the value that used in the calculation process where the data were originally in the decimal form is converted to binary. the flex sensor value in the form of binary was divided by the divisor, which is 10011. The result of this division produces a CRC value is less than 4 bits in size. The CRC value from this division is combined with the binary value of the flex sensor in which this value is called the Transmitter Frame. Transmitter frame is sent by the transmitter to the receiver. The receiver is part of the decision on the value received from the transmitter. the value of the transmitter frame received by the receiver is called the receiver frame. The scenario is that the binary value of the transmitter frame has been received at the receiver and also checked with the CRC algorithm. The receiver frame value is divided by the same divisor as in the transmitter section, then the result of the division is the CRC.

Figure 10

Figure 11 shows that the decimal to binary conversion process uses the decimal to binary function, consisting of a mathematical formula, namely binary num [i] = n% 2. This formula is an arithmetic process in C ++ or C # programming, with a residual operator often called the modulus. The modulus operator in the decimal to binary formula uses the residual value for all the divisions between n and 2, therefore it is possible that the quotient of n% 2 is 1 or 0. From decimal to binary conversion, the output of data transmitter can be described in Fig as the listing of output data.

Figure 11

Figure 11 shows the output data originating from the transmitter, with the data1 and num representing flex sensor values. The decimal value of data 1 is also converted to binary. Furthermore, the process of calculating the CRC algorithm needs a divisor or generator value with a capacity of 5 bits, namely 10011. The divisor or generator value is fixed and used as a divider. Furthermore, the divided value comprises the data 1 (binary) + dividend (appended), which is added to the back of each data1 value (binary). The dividend value is also the divisor −1, which means that it is also in 5 bits. Therefore, the dividend (appended) value is obtained by subtracting 1 from 5 for determining the value of 4 bits. the value of the transmitter frame in the decimal form received by the receiver is called the value of the receiver frame. The value of the receiver frame is divided by the same divisor as the transmitter, namely 0011. The result of the division is called a remainder. The remainder is the final result of the calculation of the CRC algorithm to determine whether the data received is correct or error that can be described in Figure 12.

Figure 12

Figure 12 shows that the CRC calculation can identify the received data without an error and also the received frame shows a correct result which the data sent corresponds to the data received with an indication of the received frame is not 0, while one of the error results can be seen in Figure 12. An error result indicated that the data sent correspond to the data received with an indication of the received frame is 0. In this research, the CRC detection was applied at the variation of distances such as 50 cm, 100 cm, 150 cm, 200 cm, 250 cm, and 300 cm. each sample was tested around 20 data. from the testing, the error detection was caused by packet loss that occurred in the communication process of nRF24L01. The error result showed that all of the value from received frame indicated with 0 value that means receiver failed to receive or to detect data sent by transmitter. From the experiment, the success percentage of the CRC detection in the electric wheelchair can be calculated by Eq 9. (9) success rate(%)=∑testing−∑test failed∑testing×100% success\,rate\left(\% \right) = {{\sum {testing - } \sum {test\,failed} } \over {\sum {testing} }} \times 100\%

In this section, error detection from a distance of 50 cm can be seen in Table 1. Data from the transmitter can be identified as an error detector in the received frame with a value of 0.

According to Table 1, the result indicates that the error value from the distance of 50 cm is 6 time out of 20 trials. Using Eq. (9), the success rate in the communication process of the nRF24L01 module at a distance of 50 cm is 70%. in the same way, the success rate in the communication process from various distances in this study can be seen in Table 2. When the flex sensor data is converted into a binary value, the cyclic redundancy check calculation.

According to Table 2, wireless communication of the smart glove as the controller of the electric wheelchair can be identified an average success rate of 65% and error detection of 35%. From these results, error detection in this system can be categorized as low category which influenced by error connection and data corrupt. The success prediction ratio was also implemented by Messiaid et al. (2021) to calculate the evaluation of web service's quality.

In general, the calculation of CRC in wireless communication is done by specifying a polynomial number as the divisor of the data to be processed which is called the divisor. Then, the calculation of the width of the divider called the highest bit position or the width of the divisor is added. For example, if the divisor is of 1001, the appended is 3. We have to make sure that the highest bit is 1. Knowing the appended appropriately is very important, because it affects the type of CRC that can be used (CRC 16, CRC 32, and others). The processed data may only be a few bits, smaller than the poly value that will be used. This will cause us not to process all the predefined poly values. To solve this problem, in basic algebraic calculations, we add a string of bits along W to the data we are going to process. The goal is to ensure that we can process all the data correctly.

In the testing of the CRC algorithm, the algorithm is compared between the system and manual calculations to obtain valid results. The validation of the CRC algorithm in this study was compared using the C++ program where the results of the CRC calculation formula were entered into the C++ program and the CRC results were calculated manually. The goal is to prove that the data results obtained from the CRC calculation in the error detection section are correct. The calculation result of the CRC algorithm for the transmitter section using C++ can be seen in Figure 13.

Figure 13

In Figure 13, it can be seen that the manual calculation result of the algorithm is 0100 which this result of manual calculation with calculations entered into the program have the same value. Furthermore, in the receiver section, the receiver frame has a value of 1011000100100 with the remainder being 0000. The following is the manual calculation result of the receiver's CRC algorithm that can be seen in Figure 14.

Figure 14

In Figure 14, it can be seen that the remainder generated from the manual calculation of the CRC algorithm is 0000, this result proves that the value of the remainder from the CRC calculation included in the program has the same value. This validation of the CRC calculation takes data samples from Table 1 number 2 of the transmitter section, namely the value 354 with a binary value of 101100010 which has a CRC result of 0100. In the division process above, we get an important thing to note in this algebraic calculation is that we do not need to perform an XOR operation when the highest bit is 0, but we only shift until the highest bit is 1. This will a little easier and speed up the algebra operation. In notation, it can be written as Eq. (10): (10) a(x).xN=b(x).p(x)+r(x) a\left(x \right).xN = b\left(x \right).p\left(x \right) + r\left(x \right) where a(x) is a polynomial number that represents data. b(x) is the value of 0 as appended. p(x) is a Divisor. r(x) is the remaining Quotient or CRC value.

From Eq. (1), it shows that r(x) is the remainder of the quotient or CRC value. the integrity data from CRC value can be done in several ways such as (1) CRC value is calculated from the original data, then match the CRC value with the stored CRC value (append to the data). From the data obtained, the CRC value is N-1. (2) The original data are added with the CRC value, divided by the divisor. Finally, the remainder of the quotient is 0 if the data are correct.

Throughput testing is carried out to determine the actual data rate (data rate) based on the unit of time. the calculating of throughput can be used in Eq. (4). throughput test scenario was carried out by sending a number of data packets from the transmitter to the receiver and the testing was carried out at 60 times with varying distance from 50 cm to 300 cm. Throughput test results can be seen in Figure 16.

The size of the data packet at every of 50 cm affects the transmission in seconds (data rate). According to Figure 16 throughput testing in varying distances of 50 cm, 100 cm, 150 cm, 200 cm, 250 cm, and 300 cm was obtained the data transmission speeds of 90.38 bps, 270.91 bps, 45.62 bps, 72.54 bps, and 44.38 bps, respectively. From the results of the analysis, it was found that throughput parameters were tested ranging from 50 cm to 300 cm which the overall average throughput was 129.38 bps. These results were compared with parameter in Table 4 in which throughput quality can be categorized as excellent and the testing results in an actual speed (throughput) of 270.91 bps. The maximum speed (bandwidth) of nrf24I01 module is 2 Mbps (2 × 103 kbps or 2 × 106 bps) and the minimum speed of the nRF24L01 module is 500 kbps (5 × 105 bps).

Jitter

The purpose of the jitter parameter is to analyze the delay variation on the transceiver of data processing on the system that causes delays from one transmission to another. The throughput test scenario was carried out by sending a number of data packets from the transmitter to the receiver and the scenario was carried out at 60 times with a varying distance from 50 cm to 300 cm. Jitter calculation is influenced by variations in delay and the amount of data received. The results of the average jitter test can be seen in Figure 17.

Figure 17

According to Figure 18 jitter value in the test of every 50 cm indicated that there is an average jitter value of 21.60 ms. These results were compared with Table 5. The average value of the jitter is good category with index of 3. Data communication between transceivers is influenced by the size of the data sent, where the larger the size of the transmitted data, the greater the jitter value obtained.

Figure 18

The testing of packet loss aims to determine the percentage of packet loss that occurs in a data communication process between the transmitter and receiver. The packet loss testing process is carried out by conducting a transceiver communication process with a varying distance of 50 cm, 100 cm, 150 cm, 200 cm, 250 cm, 300 cm. In the results of the quality of service with packet loss parameters, it can be indicated that a number of data packets have been transmitted on the communication line between the transmitter and receiver with different distance variations. The results of the packet loss test can be seen in Figure 18.

According to Figure 18, the results shows that the distance variations of packet loss were applied of 50 cm, 100 cm, 150 cm, 200 cm, 250 cm, and 300 cm with the result of 3%, 2%, 4%, 5%, 0%, and 4%, respectively. From these results, the average packet loss from distance variation is 3%. Based on Table 6, it can be categorized as good value with index of 3 in the process of transmitting data packets from the transceiver, because in the communication process the transceiver uses reliable communication for ensuring the sending of packet data by the transmitter can be received by the receiver properly even though there are some packets cannot be received by the receiver. this is caused by data packet damage or corrupt and loss of connection.