2018 Sub-1 GHz Wireless Nodes Performance Evaluation for Intelligent Greenhouse

kusumawardana@pnb.ac.id Abstract Greenhouses provide not only solution to problems faced by conventional farming systems but also play an important role to improve the energy efficiency and environmentally friendly awareness. To achieve benefits of greenhouse farming system in terms of energy efficiency, research related to this issue have been done by many researchers. However, resources that concern on how to practically implement the particular energy-saving technology for greenhouses need to be improved. In this research, field experiment results related to low-power communication between nodes have been reported by implementing universal prototype modules. The pros and cons of existing communication technology, the proposed architecture of network and module analysis, and the performance evaluation of the proposed module dedicated to intelligent greenhouse farming system were also


Network Architecture and Node Module Designs for Greenhouse System
A greenhouse needs node modules to sense and control the environmental parameters. These parameters can be soil humidity, air humidity, temperature, gasses, water level, water flow, etc. In this research, two kinds of modules were developed so-called central node and end node. The end nodes can act as a sensing node (nose), an actuating node (noac), or gateway. The noacs are generally implemented to control switches, solenoid valves, relays, etc., whereas noses are equipped with sensors to sense the environmental parameters. The IoT-based architecture for intelligent greenhouse is shown in Figure 1. Figure 1. The IoT-based architecture for greenhouse system A star-topology is implemented to realize the nodes communication. This topology offers robust network management, yet simple to be developed. The central node which acts as a main controller or gateway for sub-1 GHz protocol may consist of many end nodes. It is connected to TCP/IP or internet network through a small-sized single board computer (SBC). By using an SBC (Raspberry Pi), some services can be built, such as data logging, intelligent algorithm, user interface, etc. The implemented addressing format is similar to the IP address v4 for TCP/IP protocol. Each node has its unique device ID, network ID, and network key encrypted using an AES 128-bit.
All nodes operate wirelessly in 868 MHz frequency band. This frequency is selected as it offers clearer wave propagation compared to 2.4 GHz band which belongs to general commercial communications. This sub-1 GHz band can reach wider coverage area to support the IoT paradigm. In Figure 2, nodes installation for greenhouse is shown [17]. A node module is developed by using an 8-bit ATmega328-AU microcontroller. This microcontroller was selected as a main processor since it has all features required by both nose and noac, and it is the same chip used by Arduino UNO, one of the most popular open source platforms worldwide. Therefore, all benefits possessed by the Arduino UNO such as programming software, libraries, community supports, etc. can also be used. The hardware blocks of each module are shown in Figure 3. watch dog timer (WDT) acts to reactivate the microconcontroller from deep sleep state without additional circuits. Therefore, all modules can be reduced their power consumptions significantly.

Results and Analysis
To prove the concepts mentioned above, in this research, two prototype modules were developed to simulate both central and end nodes. Two evaluation steps were conducted. First, the transmission performance was conducted to explore the nodes characteristics. The transmission performance evaluation was conducted in two kinds of fields: at a wide seashore to find a line-of-sight (LoS) characteristics, and at a dense housing area to find non-line-of-sight (NLoS). Second, the power consumptions evaluation was conducted to find the node's life-time by employing low-powered batteries.
The transmission test was conducted by sending 50 data packets from the end node to the central node with every 300 ms interval. Powered by three NiMH AA batteries, the end node was moved every 25 meters from the central node. This process was conducted continuously until the data obtained by the central node was completely lost. By using this scenario, the maximum transmission distance can be recorded. The node module prototypes and transmission evaluation device installation are shown in Figure 4. As shown in Figure 4, the central node was connected by a USB cable to a laptop to run logger process. In Figure 5, the calculated Rmean versus transmission distance for LoS measurement is shown. According to the graph, there was a significant power decline over the first 300 meters of measurements. Then, there was a steady Rmean between -75 dBm and -85 dBm in the remaining distances and completely lost at over 90 dBm. Along middle-distance range (between 300 and 900 meters), the data packet will be potentially lost if obstacles present. The distance of 900 meters or over will hardly to achieve if the central and end node were not in a line-of-sight position. For NLoS measurement, the evaluation was conducted in a dense housing area. In order to receive more data, 50 data packets for every 15 meters range were sent from the end node to the central node, instead of 25 meters as conducted for LoS test. The received data in NLoS test is shown in Figure 6. According to the figure, a sharp decline of Rmean happened in the first 75 meters. Then, steady a Rmean between -91 dBm and -93 dBm was shown in the remaining distance and totally disappears at 135 meters. To conclude, the penetration range for NLoS is ideally up to 75 meters.

Figure 6. NLoS measurement test
As shown in Figure 6, in the worst case, where the communication blocked by many obstacles, the nodes connection can reach up to 135 meters. Assume that the greenhouse in the form of a circle area and the central node is placed in the center of the circle, in theory, the maximum greenhouse diameter will be 270 meters. If a better transmission performance is needed without considering the power consumption, a better transmission algorithm that make it possible to resend the missing packets is needed, like it does on TCP/IP protocol. Moreover, improvement in hardware parts, such as antenna design, may increase the performance.
For the life-time evaluation, the maximum and minimum electrical current consumptions (Imax and Imin) in milliamps during four processes were recorded, that are when analog-todigital-converter (ADC) is running, when sending the data, when receiving the data, and when deep-sleep mode is activated. These four processes are considered as the main activities done by the nodes. There were 100 analog data read by ADC, and duration needed to complete the conversion was recorded. Based on return value of micros() function provided by the Arduino software, the value of ADC, transmit, and receive times of the nodes in microseconds can be recorded. Therefore, by calculating the durations of each process, the average current drains of the node (Imean) can be recorded. The life-time of the node can be obtained by using formula = , where T denotes the possibility of node life-time in hours (h), BattCap denotes the battery capacity in milliAmps Hour (mAh), and Imean denotes the average current drawn in Amperes (A). based on the evaluation steps, the current drain of each process done by the node is reported in Table 3, and the current drawn is shown in Figure 7.  Figure 7. Current drawn of the node As shown in Figure 7, the current drawn increased significantly from 50 ms to 60 ms. At this period, the node is transmitting the data. Based on Table 3, the transmitting process consumes 135.60 mA. The second highest current drwan is dedicated for receiving data that is 23.12 mA. The ADC and deep-sleep mode consume 8.34 mA and 0.0041 mA, respectively. After recording the current drawn, the Imean value was calculated. The obtained Imean value during was 0.13 mA. Finally, the battery capacity in mAh was divided by Imean to obtain the lifetime of the node.
To find a general view of the node life-time, a 2550 mAh NiMH battery was used as a main power source, and calculation details are shown in Table 3. The node module can operate approximately up to 19,905 hours or 2 years and 3 months, provided that the used firmware has 0.23 duty cycle, consisting of 100 times ADC conversions, transmitting data, receiving data, and deep-sleep mode. All processes were counted only one loop (one time). The sensing node (nose) can ideally operate at 0.23 of duty cycle with only 8018 ms of full cycle time, considering that environmental parameters (temperature, humidity, light intensity, etc.) do not change immediately in 8 seconds.

Conclusion
In this research, wireless node modules operating in 868 frequency band have been developed for greenhouse system. The sub-1 GHz was selected since it offers a better wave penetration in housing area and less interferences compared to 2.4 GHz frequency band. The node architecture was developed by using an ATmega328 microcontroller, which is compatible with the Arduino open-source platform. Therefore, all benefits possessed by the Arduino such as programming software, libraries, community supports, etc. can also be used. In this research, two evaluation steps were conducted: the transmission performance test and the life-time test.
According to the evaluations, the maximum communication ranges for LoS and NLoS scenarios were 1,275 meters (Rmean -90 dBm, Ploss 98%) and 135 meters (Rmean -93 dBm, Ploss 82%), respectively. The node module can operate approximately up to 19,905 hours or 2 years and 3 months, provided that the used firmware has 0.23 duty cycle, consisting of one loop process (100 times ADC conversions, transmitting data, receiving data, and deep-sleep mode).