In an era where digitalisation is reshaping every industry, the need for robust digital solutions in outdoor challenging applications, such as agriculture, forestry, and construction, has become more crucial than ever. As we venture into these natural, often remote settings, the challenges are manifold. Rugged terrains, dynamic ecosystems, and the lack of infrastructure pose significant hurdles to digital operations. The key to overcoming these obstacles lies in establishing reliable internet connectivity, which is not just a facilitator, but a necessity for modern applications resting on the shoulders of AI-enablers, such as IoT devices and robotic solutions. 

The FEROX project emerges as a beacon in this landscape, leading the charge towards innovatively bridging the connectivity gap in a particularly challenging outdoor environments: forests. At the heart of FEROX’s mission is the integration of cutting-edge digital technologies, from autonomous drones to sophisticated data analytics, all hinging on the seamless flow of information facilitated by an advanced internet setup. This project is not just about bringing the internet to remote areas; it is about unlocking the full potential of AI and robotics in forestry, transforming how we interact with and manage these crucial ecosystems. 

By establishing a DIY remote area internet connection, FEROX is setting a precedent for digital innovation in the wilderness. This endeavour goes beyond mere connectivity; it is about creating a foundation for safer, more efficient, and sustainable forest operations, truly marking the advent of a new digital era in forestry. 

Remote Area Internet Connection Technologies

In the quest to revolutionise forest operations, the FEROX project stands at the forefront of harnessing the latest in remote area internet connection technologies. This pivotal move towards digitalisation in remote areas is not just about bringing internet connectivity to these areas; it is more about creating a seamless and integrated network that supports a wide array of advanced digital solutions. From real-time data transmission to AI-powered analytics, the foundation of all these revolutionary changes lies in robust and reliable connectivity. Central to this technological leap are two key advancements: the beyond 5G and 6G vision, as well as the role of the traditional IEEE 802.11. Both these components serve as the backbone of the project, ensuring that every drone, sensor, and other tools within the FEROX ecosystem communicates efficiently, securely, and without interruption. By delving into these technologies, we unravel how FEROX is overcoming the unique challenges presented by forest environments, setting new solutions for connectivity in remote areas. 

Individual tree identification in large scale powered by AI 

The FEROX project’s leap into forest connectivity aligns with the beyond 5G and 6G vision, signifying a major advancement in data communication, particularly for complex environments, like forests. These future networks, potentially offering data rates up to 1 Tbps, promise to revolutionize high-precision tasks, such as mapping and drone operations. Key to this vision is the reduction in latency, with 6G aiming to surpass 5G’s 1 ms target, crucial for real-time forest monitoring. The anticipated increase in area traffic capacity will also allow for the efficient sharing of detailed 3D forest maps. The expansive coverage of these advanced networks is also critical for the FEROX project, ensuring seamless communication over vast, remote forest areas. The integration of terrestrial and satellite networks in 6G is expected to enhance this coverage further. However, it brings complexities, like significant delays and Doppler shifts, requiring innovative solutions to these advanced integrated systems. 

Despite these future developments, current local infrastructure solutions, particularly IEEE 802.11 networks, remain essential. They provide reliable connectivity in dynamic forest environments, underpinning the project’s goal of seamless, efficient, and safe digital forestry operations. As beyond 5G and 6G networks develop, they will offer faster, smarter, and more adaptive capabilities, marking a significant stride in remote area connectivity for the FEROX project. 

The Role of IEEE 802.11

In transforming forest operations, IEEE 802.11, widely known as WiFi, is pivotal. This protocol suite, ranging from 802.11a/b/g/n/ac/ax, has extended wireless communication to the challenging environments of forestry, revolutionising how we connect in these remote areas. For the FEROX project, the versatility of IEEE 802.11 is indispensable; especially considering the adoption of the Robot Operating System (ROS) as a quasi-standard robotics middleware1. Different IEEE 802.11 standards cater to diverse requirements: 802.11a provides high data rates for intensive tasks, while 802.11b/g enhances coverage in expansive forests. The introduction of 802.11n, and later WiFi5 (802.11ac), marked a significant improvement in range and throughput, crucial for reliable outdoor inter-robot communication. 

This technology empowers FEROX’s UAVs and robotic systems with robust connectivity, essential in areas where cellular networks falter. IEEE 802.11 ensures seamless interaction among ROS-enabled robots, control systems, and human operators, facilitating real-time monitoring and data gathering. Its integration of TCP and UDP protocols is vital, offering the project reliable and efficient communication capabilities necessary for safe and effective forest operations. Thus, IEEE 802.11 stands as a fundamental enabler for the FEROX project’s journey into forest digitalization. 

FEROX Remote Area Internet Connection

The FEROX project’s venture into uncharted territories of forest digitalisation hinges on establishing a simple, yet reliable, DIY remote area internet connection. This sophisticated network is the linchpin in enabling seamless communication and coordination within the forest’s challenging and dynamic environment. As solutions within FEROX are all ROS-enabled, central to this network are the ROS network-related software packages to be adopted. These software packages play a unique and pivotal role, forming a cohesive and efficient network infrastructure, essential for the FEROX project’s success in remote forest connectivity. 

3.1. nimbro_network ROS package: Enhancing Robotic Communication 

The built-in network transparency of ROS creates a strong dependency on having a robust and persistent connection to the ROS master, which is unrealistic under real-world challenging scenarios, like the ones addressed in FEROX. While the most popular ROS multimaster solution, the multimaster_fkie2, tries to solve this issue by synchronising individual ROS masters on the networked hosts, leading to a distributed multi-robot solution, it does come with certain limitations, such as its reliance on ROS master handshake mechanisms, the inability to automatically discover remote hosts and topics, and the lack of advanced network features, such as transparent compression or forward error correction.  

This is where the integration of the nimbro_network ROS transport package becomes vital. Originally developed for the DLR SpaceBotCup competition and further refined by the Czech Technical University for multi-UAV solutions, the nimbro_network is tailored for reliable data transmission in challenging network environments. It significantly enhances the current ROS network stacks’ capabilities by providing robust communication tools designed to withstand adverse network conditions (Table 1). The nimbro_network software package complements the ROS stack by addressing its limitations, particularly in high-latency or unstable network scenarios, common in forest areas. It includes advanced features, like transparent compression and forward error correction, adding layers of resilience and efficiency to data transmission. This integration ensures that the communication between UAVs, ground control stations, and human operators remains uninterrupted and efficient, which is crucial for real-time tasks and mission-critical data exchange.  

Table 1. Comparison between the regular ROS network transparency, the multimaster_fkie package, and the nimbro_network package. 

3.2. ros-network-analysis ROS package: Assessing Network Performance 

In ensuring the efficacy of the FEROX project’s remote area internet connectivity, the ros-network-analysis ROS package plays a pivotal role in evaluating and optimizing network performance3. This tool is essential for navigating the dynamic and demanding conditions encountered in forest environments, where robust and reliable communication forms the backbone of effective data exchange and coordination among various devices. This package delves deep into key performance metrics, such as signal quality, network latency, throughput, and connection errors4 (Table 2). 

Table 2. Output data description of the ros-network-analysis ROS package. 

Parameter 

Description 

Network Quality 

The Received Signal Strength Indicator (RSSI) measures the power level within a received wireless signal in decibels (dBm). It gauges the quality of the network signal, providing insights into signal strength. To accomplish this, the method scans the specified interface name, reading the /proc/net/wireless. 

Network Throughput 

This metric quantifies network throughput in megabits per second (Mbps). It captures data related to transmitted and received data packets and data rates (in Mbps) for both TCP and UDP connections. To gather this data, the node utilises the /proc/net/dev file to identify the interface name and extracts the necessary information from the /proc/net/snmp file for each TCP and UDP connection. 

Network Errors 

Network error metrics are documented, including the total number of data for categories such as retransmitted packets and retries. These metrics offer insights into the reliability of data transmission and are obtained from sources such as netstat and ethtool. Additionally, it utilises ethtool to fetch data on rx_dropped and tx_retries. In the process, the method reads the file /sys/class/net/" + interface_name +"/statistics/tx_errors (or tx_dropped, rx_errors, or rx_dropped) to gather the necessary data, which is subsequently published. 

Network Delay 

Network delay, measured in milliseconds (ms), records application-level network round-trip time latency. It also notes the network's connectivity status, indicating whether it is currently active or temporarily disconnected. In cases of network inactivity, the recorded delay value is set to -1 ms. 

3.3. DIY Setup: Hardware Components 

Embarking on a DIY project for establishing remote area internet connectivity, especially in the context of forest operations like those in the FEROX project, necessitates a carefully selected bill of materials (BoM). The hardware components presented in this section serve a dual purpose: they are crucial for the initial network evaluation and also provide a foundational guide for further customisation based on specific needs. It is noteworthy that, while these components are pivotal for evaluating the network’s performance in a forest setting, it is equally important for readers to consider their unique requirements when tailoring their own BoM. This adaptability is key to successfully implementing a reliable internet connection in challenging outdoor environments. 

Figure 1. Initial setup at start point position. 

In the FEROX project, the cornerstone of the remote area internet setup is the NFT Blizzard 2ac-N access point (AP)5 designed for harsh environments, illustrated as the WiFi BaseStation in our setup in Figure 1. With dual 2.4/5GHz 2×2 MIMO radios, it delivers a 1.167Gbps data rate and 29dBm output power, ensuring wide coverage and reliable connectivity. Its metal casing provides noise immunity, heat dissipation, and durability, while compliance with IEC 61000-4-5 standards ensures resilience in adverse weather conditions. Network meshing capabilities further enhance its utility for very large environments. If internet connectivity is required, and in the expectation that the BaseStation will be in cellular communication coverage, then a router can also be added. We have conducted experiments with the ZBT CPE2801 rugged 4G router6, though any PoE-enabled 4G or 5G router would work. 

For network evaluation, we used two laptops - a Dell-XPS15 and an Acer A515, both equipped with advanced Wi-Fi capabilities and running Ubuntu 20.04 with ROS Noetic Ninjemys. These laptops were used to manage the data flow in the network by resorting to ROS bags previously acquired by drones in the forest, containing vital data like images and point clouds, simulating the network load under real-world settings. Also for network performance assessment purposes, two real-time kinematics (RTK) global navigation satellite systems (GNSS) receivers. For this purpose, the REACH RS2+7 solution was chosen, although similar alternatives could also be applicable. The REACH RS2+ is a multi-band RTK GNSS receiver offering centimetre-level precision (though not under the forestry canopy), essential for surveying, mapping, and navigation. Its compatibility with various satellite systems allows for precise and adaptable positioning data collection, crucial for the accurate evaluation of the FEROX project’s network infrastructure. In our setup, the primary role of the two RTK GNSS devices was to measure the distance between WiFi-enabled devices, namely the laptops mentioned earlier. This distance measurement was pivotal in understanding the correlation between range and network performance metrics. 

At last, but not least, in the FEROX project, the use of batteries is a vital aspect of powering the network’s key components, especially in remote forest environments where traditional power sources are unavailable. Common LiPo batteries, typically used in drones, provide an adaptable and portable power solution. These batteries, offering 26v, can be efficiently converted to the required 48v for Power over Ethernet (PoE) components, like the AP and the router, using a standard DC-DC step-up converter. This setup ensures continuous and reliable operation of the access points, crucial for maintaining uninterrupted network connectivity. If money is not the issue, then the alternatives fall on the use of specific field battery banks, like FlashFish solutions8, though DC-D step-ups to 48v may still be required. 

Concluding this section on the DIY setup for remote area internet connectivity, the FEROX project underscores the importance of a meticulously curated BoM, tailored to your specific needs. This setup not only serves as a blueprint for successful forest connectivity, but also offers valuable insights for similar projects seeking to establish reliable communication in other remote and demanding environments.  

3.4. A Glimpse at the Network Performance: Preliminary Insights 

In this section, we offer a sneak peek into the preliminary findings obtained from the FEROX project’s network setup deployed in the forest. While a detailed analysis and comprehensive results will be presented in an upcoming technical report, here we briefly explore how key network metrics, specifically the network quality and delay, vary with distance. This analysis focuses on the relationship between these metrics and the spatial separation between our portable device (Dell XPS15) and the BaseStation laptop (Acer A515).  

Figure 2. Box plots of RSSI and throughput over the spatial separation between WiFi devices. 

The box plots in Figure 2 depict the RSSI and throughput over distance, providing compelling insights into network performance in challenging outdoor environments. As expected, both RSSI and throughput exhibit a gradual decline as the mobile device moves farther away from the AP. However, the results reveal an encouraging and noteworthy trend - consistent throughput is maintained up to the 140-meter mark, with data still being exchanged beyond that point, albeit with reduced reliability. In practical terms, this implies that with the suggested setup, robotic systems operating within the FEROX project’s forest environment would still be capable of efficiently sharing substantial data, including large point clouds and critical information, even when operating at considerable distances of up to 140 metres from the central AP (i.e., ~6 hectare around the AP). Furthermore, and since the proposed setup is compatible with mesh networks, then integrating additional APs would significantly increase area coverage. This finding underscores the robustness and viability of the network infrastructure, offering a promising foundation for deploying digital solutions in dynamic and unstructured outdoor landscapes.