Unmanned Aerial Vehicles (UAVs), otherwise known as drones, are aircraft devices which operate without an onboard pilot. The navigational autonomy of UAVs previously enabled applications for military, telecommunications and surveillance, however, only recently have the communication capabilities of these devices been considered.
Among balloons and small aircraft, UAVs offer reliable and cost-effective wireless communication solutions (Bucaille et al, 2013). Using a low altitude platform (LAP), such as UAVs, allows for easy deployment and line-of-sight (LoS) communication links, unlike high altitude platform (HAP) alternatives, such as balloons. However, HAPs offer increased endurance and coverage, ergo are more suitable for covering large geographic areas.
Zeng et al. (2016, p. 37) introduced three potential aids UAVs could provide: ubiquitous coverage, relaying and information dissemination and data collection. UAV-aided ubiquitous coverage refers to the deployment of UAVs to assist in providing consistent coverage across the current communication infrastructure. For instance, base-station offloading in extremely crowded areas, one of the key five scenarios outlined to be addressed by fifth-generation (5G) systems (Osseiran et al., 2014). UAV-aided relaying is the process of deploying UAVs where establishing reliable communication links between parties is difficult, or impossible. An example being positioned between the front-line and command centres. UAV-aided information dissemination and data collection is an unmanned aerial system (UAS), consisting of UAVs which accumulate delay-tolerant information from a large amount of other distributed wireless devices. For example, wireless sensors in precision agriculture applications (Zeng & Zhang, 2017; Gholami, 2019.).
Figure 1 depicts two network examples: a single UAV network or a multi-UAV network. Multiple UAVs within a UASs create a mobile ad-hoc network (MANET), however, UAV networks are also an example of vehicular ad-hoc networks (VANETs). Sahingoz (2013) explains how the “highly mobile” characteristic of this network differs from typical ad-hoc networks, resulting in a new networking model referred to as airborne networks, unmanned aeronautical ad-hoc networks (UAANETs) or, more commonly, flying ad-hoc networks (FANETs). Khan et al. (2020) detail the benefits of FANETs, rather than a single UAV-assisted network, in regards to network scalability, survivability and payload distribution.
Zhao et al. (2018) compared MANETs to traditional wireless networks, outlining the key differences between the Open Systems Interconnection (OSI) communication model and the new technologies: the physical, link and network layers. This is demonstrated in Figure 2. FANETs have sophisticated mobility compared to MANETs, but use the same networking structure and wireless communication links, such as IEEE 802.11 a/b/g/n (Han et al., 2009).
The physical layer uses basic signal transmission technologies via antennas by modulating bits into sinusoidal waveforms. An example is the radio propagation model, which uses electromagnetic waves, but communication links decrease in quality over time. The radio propagation model as well as antenna structure have been development points for researchers as a result. However, Ahmed et al. (2011) noted many studies had been conducted in 2D environments, in which antennas behave differently compared with the 3D topologies of FANETs. In the same way, there is yet to be a transport layer protocol introduced for FANETs. The first FANET systems used existing transport protocols, supporting Transmission Control Protocol (TCP) and User Datagram Protocol (UDP). However, TCP performs poorly in both MANET and FANET environments. A substitute protocol, designed for UAVs, is STANAG 4586; this protocol aimed to support interoperability across multiple UAVs and ground control stations (GCSs), but is not specialised for FANETs (Cummings et al., 2006).
Similar to terrestrial networks, FANETs require communication links and routing. However, control and non-payload communication (CNPC) links are additionally required for supporting safety-critical functions of UAVs, such as collision and crash avoidance. UAVs therefore consist of two basic types of communication links: the CNPC link and the data link. Alternatively, the network layer inherits static, proactive and reactive routing protocols from MANETs.
The CNPC Link is designed to ensure safe operation of UAVs, by enabling highly reliable and secure two-way communications. These communications may be: commands from a ground control station (GCS) to UAVs, the uplink; aircraft status reports from UAVs to the GCSs, a downlink; or sense-and-avoid information among UAVs. An example of a GCS would be a dedicated mobile terminal on the ground. A backup beyond-radio-LoS CNPC link via satellite can also be introduced to enhance reliability and robustness, despite the preference for direct links between UAVs and GCS to prevent delays.
The Radiocommunication Sector of the International Telecommunications Union (ITU-R) identified the command and control communication LoS bandwidth requirement for UAVs as 34 MHz, and beyond-LoS as 56MHz (ITU-R, 2009). LoS CNPC links operate within either the L-band or the C-band set of frequencies (Zeng et al., 2016; Kerczewski, 2017; Matolak & Sun, 2017.), as allocated at the World Radiocommunication Conference (El-sheikh, 2020.). However, Tan et al. (2020) noted spectrum scarcity since these frequency bands are particularly congested across the globe.
The data link layer provides support for application-related communication with ground terminals, and ergo must allow for the following communication modes:
- Direct mobile-to-UAV communication
- UAV-to-Base Station and UAV-to-Gateway wireless backhaul
- UAV-to-UAV wireless backhaul
The two channels UAVs use to communicate are: air-to-air and air-to-ground (A2G). Both channels rely on LoS connections, although more so air-to-air, which are susceptible to interference due to the complex operating environment (Gholami, 2019.). UAV manoeuvring results in airframe shadowing while reflection, scattering and diffraction caused by surroundings, otherwise known as multipath propagation, fade the signal (Li et al., 2015; Sun & Matolak, 2015).
Static routing protocols pre-compute routing tables prior to UAV deployment. An example of static routing is load, carry and deliver (LCAD) where a UAV receives information from a GCS and flies to the destination GCS for the delivery (Cheng et al., 2007). Alternatively, multi-level hierarchical routing is used in accordance with many UAVs organised into swarms, with a designated node for communicating up and down hierarchy layers, potentially to GCSs and satellites (Sun et al., 2011). Static routing protocols are most appropriate for UAV-aided information dissemination and data collection.
Proactive routing protocols periodically refresh routing tables and update the information of neighbouring UAVs. Optimized link state routing, a proactive link-state protocol, is an example which synchronises the network by directly sending routing information to adjacent nodes (Clausen & Jacquet, 2003).
Reactive routing protocols do not determine routing paths until they are required, addressing the overhead induced by proactive routing. Ad-hoc on-demand distance vector (AODV) routing, an instance of a reactive protocol, discovers routes, transmits packets and maintains any routes as long as they are required (Perkins & Royer, 1999; Nayyar, 2018.).
There are two types of UAV: fixed-wing and rotary-wing. Compared to rotary-wing UAVs, fixed-wing UAVs such as small aircraft have more weights, higher speed, and they need to move forward in order to remain aloft. Rotary-wing UAVs are ergo more suited to LCAD UAV-aided information dissemination and data collection. In contrast, rotary-wing UAVs, such as quadcopter drones, can hover over a particular area for BS offloading and capabilities. The large variation of UAVs available within these umbrella families allow for flexibility in network design, capabilities and cost.
There are however constraints with UAVs by size, weight, and power (SWAP). The high mobility environments UASs are generally utilised in cause irregular connectivity and, as previously mentioned, Li et al. (2015) details natural multipath propagation, additional limitations in communication, computation, and endurance capabilities would further hinder network potential. In the same way, there are no fixed backhaul links or centralised control. UAV-assisted networking is therefore more challenging than the terrestrial alternatives.
Path planning is the process of a UAV determining the best path between its present location and destination. Han et al. (2009) studied connectivity improvement when supporting a MANET with a UAV and, for each form of connectivity explored, enhancement was consistently achieved through movement optimisation. However, FANETs and network applications also have to be considered where one or multiple UAVs could be stationary or mobile and susceptible to terrain interference. Path planning algorithms ergo important for efficient trajectory calculations, hence are continuously improved.
An early study (Bortoff, 2000) focuses on UAV path planning in hostile territories, ultimately introducing the two-step method of map generation then adding additional virtual forces. This concluded in an algorithm suitable for real-time implementation with a single UAV, but required compromise between a short or stealthy path. More recently, Aggarwal and Kumar (2020) considered artificial intelligence (AI) techniques for UAV path planning. One instance involves individual UAVs observing the mobility patterns of others in the FANET to determine it’s route (Zheng, 2018), another, designed by Pan et al. (2017), uses heuristic-search techniques. (Challite et al., 2019).
Generic issues algorithms face, include: the infinite number of variables, due to the continuous UAV trajectory to be determined; and practical constraints, such as connectivity, fuel limitation, collision, and terrain avoidance (Zeng et al., 2016; Khan et al., 2020).
Due to SWAP considerations, all UAV-assisted networks must be able to perform with the limited onboard energy. Energy-aware deployment ensures the overall network is not impacted (heavily) when a UAV requires energy replenishment, whereas energy-aware operations manage energy to consume the minimum amount required only. An example of energy-aware deployment is sequential energy replenishment so no more than one UAV needs recharging at once. Avoiding unnecessary aircraft maneuvers is an instance of energy-aware operation, however, Filippone (2006) determined hovering with zero speed to be inefficient energy consumption. (Zeng et al., 2016; Zeng & Zhang, 2017.).
The direct mobile-to-UAV communication, enabled by the data link layer, allows for base station (BS) offloading and UAV recovery during complete BS malfunction. This is carried out by directly connecting nearby mobile terminals for device-to-device (D2D) communication, an effective capacity improvement technique in terrestrial communication systems; the current use of interference mitigation and spectrum sharing can be directly applied in UAV-aided communications (Asadi et al., 2014).
Similarly, UAVs can be developed into a BSs to create flexible and on-demand UAV-based aerial networks (Merwaday & Guvenc, 2015; Gholami, 2019.) – a temporary manner of UAV-aided ubiquitous coverage, mentioned by Zeng et al. (2016). This is further explored by Mozaffari et al. (2019) suggesting a “heterogeneous 5G environment” in which LAP-UAV-Bss service temporary events and HAP-UAVs provide coverage in rural areas, however, the aerial networks proposed by Merwaday and Guvenc (2015) could be considered as more essential since they’re designed for natural disaster recovery.
The 3rd Generation Partnership Project (2019), a number of mobile telecommunication standard organisations, continue to develop requirements for UAVs in 5G environments, alongside D2D, millimetre wave (mmW), and multiple-input multiple-output (MIMO) communications.
Challenges and Considerations
As outlined throughout the report, there are challenges that must be addressed:
- Despite the STANAG alternative to TCP for UAV communications, there is no dedicated transport layer protocol designed for FANETs and ergo a dedicated protocol (or several) need to be introduced.
- Khan et al. (2020) highlights three key issues for FANET routing: highly dynamic network topology, high costs and low residual energy. Dedicated routing protocols for FANETs need to be developed in the future with energy-aware operation as well as high throughput and low packet loss.
- The frequency bands for the CNPC links were identified as congested by Tan et al. (2020), therefore, as UAV-assisted networking and communication is introduced, ITU-R will need to modify band allocations.
- Path planning with multiple UAVs is imperative for large-scale FANETs, ergo further refinement of AI in routing algorithms should be a priority for optimising communication (Challita et al., 2019).
- On top of these, national regulations must be taken into account. The Civil Aviation Authority is the governing body in the UK and the Federal Aviation Administration is the US equivalent, and these regulate UAV usage. Ergo, network designs must abide to these regulations and adapt in different countries. A main concern, in regards to current regulations, is managing UASs alongside manned planes. Alturbeh & Whidborne (2020) proposed a decision-making system for both UAVs and manned aircraft to initiate collision avoidance manoeuvres. However, this is still an issue to be tackled.
- Additionally, it is critical further studies, as highlighted by Ahmed et al. (2011), should be conducted in 3D environments to reflect the realistic conditions UASs will experience.
This report presents a review of UAV-assisted networking and communication architecture, design and applications. Basic communication architecture has been introduced and detailed networking architecture explained. Moreover, design considerations for promising UAS applications have been discussed with potential challenges for the future. Some performance-enhancing techniques have been additionally suggested. It is hoped the described challenges and applications will assist future researchers in the design and development of UAV-enhanced wireless communication systems.