The rapid miniaturization and cost reduction of Unmanned Aerial Vehicles (UAVs), or UAV drones, have propelled UAV networks into the spotlight for their exceptional capability to enable robust multi-drone collaboration. In the centralized network architecture we focus on, a central control node orchestrates the entire network, offering significant advantages in state monitoring, task allocation, and data routing over distributed architectures. This unified scheduling approach minimizes conflicts and redundancy, thereby enhancing network stability and collaborative efficiency. For many collaborative missions executed by multiple UAV drones, the central node often does not need to broadcast messages to all participants. Instead, it can efficiently disseminate commands to specific multicast groups. Consequently, designing a reliable multicast communication protocol with low latency and minimal conflicts is paramount. However, traditional Time Division Multiple Access (TDMA) protocols are typically optimized for communication scenarios involving only a single antenna mode, with their frame synchronization and slot allocation mechanisms designed accordingly. With the advancement of smart antenna and adaptive antenna technology, communication systems now require the ability to dynamically switch between different antenna modes to adapt to diverse channel conditions and service requirements.
Smart antennas generally operate in two primary modes: Omnidirectional Transmit Directional Receive (OTDR) and Directional Transmit Omnidirectional Receive (DTOR). While DTOR mode, where nodes listen omnidirectionally, allows for immediate channel sensing without prior knowledge of neighbor positions—thus reducing neighbor discovery latency—it suffers from prolonged global scan times, making it unsuitable for rapidly changing network topologies. Other approaches, such as using dual frequency bands for neighbor discovery and data transmission or employing two independent transceivers, introduce hardware complexity, increased cost, and spectrum inefficiency. Furthermore, existing hybrid antenna TDMA protocols, like the Centralized UAV Media Access Control Protocol (CU-MAC), often rely on fixed frame structures. These structures struggle to adapt to dynamic traffic loads, leading to inefficient slot utilization under low traffic and potential delays under high traffic. They also primarily address unicast and broadcast scenarios, leaving multicast communication inadequately optimized for networks of UAV drones.
To address these critical gaps, this work introduces the Sector-based Collaborative Optimized Protocol for Hybrid Antenna Mode TDMA (SCOP-HTDMA). Our protocol is designed for centralized networks of UAV drones where a central node communicates with several subordinate nodes, all equipped with identical omnidirectional/directional dual-mode smart antennas. The core contributions of SCOP-HTDMA are threefold, specifically tailored to enhance the performance of networks comprising UAV drones:
- Geomagnetic Field-Assisted Sector Confirmation Mechanism: We propose a novel mechanism that enables intra-frame synchronization in DTOR mode without relying on prior GPS-based location and timing information. By leveraging the Earth’s magnetic field as a common directional reference and utilizing binary sector encoding along with sector symmetry, nodes can accurately determine transmitting and receiving sectors. This provides a new pathway for clock synchronization in scenarios where GPS signals are compromised or unavailable for UAV drones.
- Traffic-Adaptive Dynamic Frame Structure Optimization: Moving beyond rigid frame designs, we introduce an optimization scheme that dynamically adjusts the superframe length based on the actual traffic load. This approach prevents slot wastage during periods of low activity and helps contain delays under high traffic conditions, thereby improving overall slot utilization for the communication needs of UAV drones.
- Sector-Aware Omnidirectional-Directional Collaborative Multicast Mechanism: We design an intelligent multicast strategy that judiciously chooses between omnidirectional and directional transmission modes—or a combination of both—based on the spatial distribution of target nodes relative to the central UAV drone. This mechanism minimizes slot overhead and reduces latency for multicast traffic, a common requirement in coordinated missions involving multiple UAV drones.
Network Model and Problem Formulation for UAV Drones
We consider a centralized network model highly relevant for controlled fleets of UAV drones. The network consists of one central UAV node and up to \(N\) subordinate service UAV nodes (where \(N \leq 29\) in our study). Each UAV drone is equipped with an identical smart antenna capable of switching between two modes:
- Omnidirectional Mode: The antenna radiates power uniformly in all directions, providing broad coverage with a radius denoted as \(D_o\).
- Directional Mode: The antenna concentrates energy into a specific beam with an azimuthal beamwidth of \(60^\circ\), offering a longer communication range with a radius \(D_d\), where typically \(D_d > D_o\). In our model, we set \(D_d = 2 \times D_o\).
The primary challenges identified in existing protocols for networks of UAV drones are:
- Intra-frame Synchronization Difficulty in DTOR Mode: Without GPS-provided location and timing data, a node receiving a directionally transmitted beacon omnidirectionally cannot determine the precise sector from which the transmission originated. This ambiguity prevents it from sending a directional acknowledgment, breaking the synchronization handshake.
- Inefficient Slot Utilization due to Fixed Frame Structure: Fixed-length frames cannot adapt to varying traffic patterns common in UAV drone operations, leading to idle slots (low utilization) or excessive queuing delays.
- Suboptimal Multicast Delivery: Using only omnidirectional multicast fails to reach nodes outside \(D_o\) but within \(D_d\). Using only directional multicast may require scanning multiple sectors, wasting slots if target nodes are clustered or within omnidirectional range.
Design of the SCOP-HTDMA Protocol for UAV Drones
The SCOP-HTDMA protocol is engineered to address the aforementioned problems systematically for networks of UAV drones.
Frame Structure for UAV Drones
The protocol operates with a dynamic superframe structure divided into two main phases: Neighbor Discovery and Data Transmission. Unlike static frames, the length \(T_f\) of a superframe in SCOP-HTDMA is adaptive:
$$ T_f = \min(T_{\text{total}}, 100 \text{ ms}) $$
where \(T_{\text{total}}\) is the sum of all allocated slots (fixed, contention, dynamic, and broadcast). An upper bound of 100 ms ensures bounded latency for critical traffic from UAV drones. The first superframe is dedicated to network entry, comprising an Omnidirectional Neighbor Discovery phase followed by a Directional Neighbor Discovery phase. Subsequent superframes begin with a directional broadcast of the slot allocation map (beacon) by the central UAV drone.
Hybrid Antenna Mode Operation for UAV Drones
1. Geomagnetic Field-Assisted Sector Confirmation Mechanism
This mechanism enables synchronization in DTOR mode without GPS. All UAV drones use the Earth’s magnetic field to align their local coordinate systems, defining North as a common reference (\(0^\circ\)). The \(360^\circ\) plane around each node is divided into \(K\) sectors (e.g., \(K=8\)), each assigned a unique binary code. The mapping for \(K=8\) is shown in Table 1.
| Azimuth | Sector Code | Azimuth | Sector Code |
|---|---|---|---|
| North | 000 | South | 100 |
| Northeast | 001 | Southwest | 101 |
| East | 010 | West | 110 |
| Southeast | 011 | Northwest | 111 |
When a transmitter UAV drone (Node A) sends a beacon directionally from its sector with code \(C_A\) (e.g., 001), it includes \(C_A\) in the packet. The receiver UAV drone (Node B), operating in omnidirectional mode, receives the beacon. To determine the correct sector to use for its directional reply (i.e., the sector pointing back to Node A), Node B simply computes:
$$ C_{\text{reply}} = \text{NOT}(C_A[0]) \| C_A[1:] $$
where NOT denotes bit inversion of the Most Significant Bit (MSB). This works due to the sector symmetry inherent in the coordinate system aligned to geomagnetic North. For the 8-sector case, this is equivalent to toggling the first bit, which corresponds to a \(180^\circ\) symmetry. This allows Node B to send an acknowledgment back to Node A directionally, completing the handshake and enabling synchronization for the network of UAV drones.
2. Traffic-Adaptive Frame Structure and Length Optimization
The superframe combines fixed assigned slots (for basic control and guaranteed traffic from UAV drones) and dynamically assigned slots. The central UAV drone calculates \(T_{\text{total}}\) based on current allocations and sets the next superframe length accordingly. This dynamic adjustment significantly improves slot utilization compared to fixed-length frames, directly benefiting the efficient operation of UAV drones.
3. Omnidirectional-Directional Collaborative Multicast Mechanism
This mechanism optimizes multicast delivery from the central UAV drone to a group of subordinate UAV drones. Let \( \mathcal{T} \) be the set of target nodes, \( \mathcal{N}_o \) be the central node’s omnidirectional neighbor table, and \( \mathcal{T}_o = \mathcal{T} \cap \mathcal{N}_o \). Let \( \mathcal{T}_d \) be the targets outside \(D_o\) but within \(D_d\), distributed across \(n\) distinct directional sectors. The central node decides the transmission strategy as follows:
- If \( \mathcal{T} \subseteq \mathcal{N}_o \), use a single omnidirectional transmission.
- Else, calculate the slot cost for two strategies:
- Collaborative Multicast Cost (\(C_c\)): One omnidirectional slot for \( \mathcal{T}_o \) plus \(n\) directional slots (one per unique sector for nodes in \( \mathcal{T}_d \)), so \(C_c = 1 + n\).
- Pure Directional Multicast Cost (\(C_d\)): This requires broadcasting in all \(K\) sectors to ensure coverage, or at least in the \(m\) sectors where *any* target node resides. In the worst case, \(C_d = K\). A more optimized directional scan would target only the \(m\) sectors containing any target node, so \(C_d = m\). Typically, \(m \geq n+1\) if some targets are in \( \mathcal{N}_o \).
- Choose the strategy with the lower slot cost. The decision function is:
$$ C_{\text{multicast}} =
\begin{cases}
1, & \text{if } \mathcal{T} \subseteq \mathcal{N}_o \\
\min(1 + n, \, m), & \text{otherwise}
\end{cases} $$
This intelligent selection minimizes multicast latency and slot consumption for commands sent to groups of UAV drones.
Performance Analysis for Networks of UAV Drones
We compare SCOP-HTDMA theoretically against PPMAC and CU-MAC protocols in a scenario with 20 UAV drones and a basic packet inter-arrival time of 20 ms.
Average Delay: For SCOP-HTDMA, with a typical calculated \(T_{\text{total}} \approx 13.5 \text{ ms}\), the average packet delay \(D_s\) can be approximated as half the superframe length due to uniform arrivals:
$$ D_s \approx \frac{T_f}{2} \approx 6.75 \text{ ms} $$
For CU-MAC with a fixed frame, packets arriving just after their assigned slot wait nearly an entire frame, leading to a significantly higher average delay \(D_c \approx 56 \text{ ms}\). PPMAC, based on contention, suffers from unpredictable delays that increase with collisions.
Multicast Efficiency: Consider a multicast to nodes located in \(m\) sectors. A pure directional protocol like CU-MAC requires \(m\) slots. SCOP-HTDMA’s collaborative mechanism reduces this cost to \(1 + n\), where \(n \leq m-1\). When all target UAV drones are within omnidirectional range (\( \mathcal{T} \subseteq \mathcal{N}_o \)), the cost is just 1 slot, maximizing efficiency.
Simulation Results and Discussion for UAV Drones
We implemented SCOP-HTDMA in OPNET and compared it with PPMAC and CU-MAC. Key simulation parameters for the UAV drone network are listed in Table 2.
| Parameter | Value | Parameter | Value |
|---|---|---|---|
| Scene Size | 5 km × 5 km | Data Rate | 6 Mbps |
| Basic Packet Size | 128 Bytes | Video Packet Size | 64-2048 Bytes |
| Mobility Model | Smooth Motion | \(D_o / D_d\) | 0.5 km / 1 km |
| Basic Packet Rate | 10 packet/s | Video Packet Rate | 0-300 packet/s |
The results clearly demonstrate the advantages of SCOP-HTDMA for networks of UAV drones:
- Packet Loss Rate vs. Number of UAV Drones: As the number of UAV drones increases, PPMAC’s loss rate rises sharply due to increased contention collisions. Both CU-MAC and SCOP-HTDMA maintain low loss rates thanks to conflict-free TDMA scheduling. SCOP-HTDMA, using directional transmission for data, achieves marginally lower loss than CU-MAC by reducing interference.
- Average Delay vs. Number of UAV Drones: Delay increases with network size for all protocols. However, SCOP-HTDMA consistently exhibits the lowest average delay. The dynamic frame adjustment prevents the long queuing delays inherent in CU-MAC’s fixed frame structure and avoids the collision-induced delays of PPMAC, making it highly suitable for real-time applications involving UAV drones.
- Average Delay vs. Packet Size (Traffic Load): Under low traffic loads, the adaptive frame length of SCOP-HTDMA keeps delays minimal. CU-MAC’s fixed frame introduces unnecessary idle time, increasing delay. PPMAC shows variable performance due to contention.
- Multicast Average Delay vs. Node Distribution: As multicast target UAV drones are spread across more sectors, the delay for CU-MAC increases linearly. SCOP-HTDMA’s collaborative multicast mechanism shows a much slower increase in delay because it efficiently uses a single omnidirectional transmission for in-range nodes, supplementing with only the necessary directional transmissions.
Conclusion
In this work, we presented SCOP-HTDMA, a dynamic scheduling TDMA protocol designed to overcome key limitations in existing hybrid antenna MAC protocols for centralized networks of UAV drones. By introducing a geomagnetic-assisted sector confirmation mechanism, we enabled robust intra-frame synchronization in DTOR mode without dependence on external GPS signals. Our traffic-adaptive dynamic frame structure optimizes slot utilization, effectively reducing average packet delay across varying load conditions. Furthermore, the sector-aware omnidirectional-directional collaborative multicast mechanism minimizes resource consumption and latency for group communications, a frequent requirement in UAV drone team operations. Simulation results confirm that SCOP-HTDMA outperforms benchmark protocols like PPMAC and CU-MAC in terms of packet loss rate, average delay, and multicast efficiency, making it a compelling solution for reliable and efficient communication in collaborative UAV drone networks. Future work will focus on enhancing the initial network entry process to reduce collision probability as the scale of UAV drone networks expands further.