In recent years, Unmanned Aerial Vehicles (UAVs or drones) have seen rapidly expanding applications across various sectors in China. These include collaborative military operations, search and rescue missions, ecological environment monitoring, and modern agricultural practices. A UAV Ad-Hoc Network (UANET), formed by multiple drones working cooperatively, offers significant advantages over a single drone system. It provides broader coverage, greater operational flexibility, and enhanced robustness through distributed collaboration. Within the clustered architecture of a UANET, the two-hop clustering structure is particularly promising for its scalability, as it generates fewer cluster heads compared to single-hop structures. This architectural choice, however, imposes specific demands on the design of efficient intra-cluster routing protocols.
Routing protocols for ad-hoc networks are broadly categorized into proactive, reactive, and hybrid types. Proactive protocols, like the Optimized Link State Routing (OLSR) protocol, maintain routing information through periodic control messages, offering low communication latency at the cost of higher control overhead. Reactive protocols, such as AODV, discover routes on-demand, reducing overhead but increasing latency. Hybrid protocols like ZRP attempt to balance these trade-offs. For China UAV drone networks, which often have stringent low-latency requirements, proactive protocols like OLSR hold considerable research value, especially in scenarios with limited topological scope where routing can be established using simple neighbor discovery messages without extensive control flooding.
Existing efforts to reduce the control overhead in OLSR primarily focus on two strategies: adaptive message intervals and cross-layer collaboration. The former dynamically adjusts the frequency of control messages like HELLO based on network stability, but suffers from topology perception delays. The latter leverages information from other protocol layers (e.g., data priority, link quality, node mobility) to inform routing decisions, potentially avoiding redundancy but sometimes introducing its own overhead. These approaches have not been fully optimized for the specific constraints of two-hop clustered networks common in China UAV drone operations. To address this gap and achieve a better balance between control overhead and network responsiveness, this paper proposes a novel Cross-Layer Collaborative OLSR protocol, named CLC-OLSR, specifically designed for two-hop China UAV drone ad-hoc networks.
System Model and Protocol Framework
The network topology targeted by CLC-OLSR is a centralized two-hop cluster. At the data link layer, a central node (functioning as the cluster head) manages network coordination, time slot allocation, and synchronization. All other nodes are ordinary members and communicate under the central node’s scheduling. Crucially, every ordinary node is within the two-hop communication range of the central node, forming a logical two-hop neighborhood. The data link layer employs a centralized Two-Hop Time Division Multiple Access (TDMA) protocol, enabling efficient, collision-free transmissions and flexible resource allocation.
In contrast, at the network layer, all nodes have equal status in a flat, peer-to-peer structure. They run the same routing protocol, adhering to unified rules for data forwarding and topology maintenance. The central node’s managerial role at the data link layer does not grant it special privileges at the network layer. This paper assumes a stable, pre-existing logical cluster and focuses on optimizing the data plane performance of intra-cluster routing, leaving initial cluster formation and head election processes outside its current scope.
Design of the CLC-OLSR Protocol
The CLC-OLSR protocol is built upon the standard OLSR framework but introduces deep cross-layer synergy with the underlying Two-Hop TDMA data link layer. This synergy is realized through three key mechanisms designed to minimize control overhead while maintaining rapid route convergence.
1. Asymmetric Link Optimization Mechanism
Traditional OLSR requires multiple rounds of HELLO message exchanges to establish a symmetric link between two nodes. Furthermore, asymmetric neighbors that never become symmetric persist in HELLO messages, accumulating control overhead. CLC-OLSR introduces an Asymmetric Link Optimization mechanism that reuses adjacency information naturally acquired by the data link layer.
In the underlying TDMA protocol, if node A receives a Beacon or network entry request message from node B, the data link layer can deduce that B is within A’s communication range. This relationship is topologically equivalent to an asymmetric link at the network layer. This information is passed upward, allowing the network layer to immediately populate its asymmetric neighbor list without waiting for HELLO exchanges. This pre-population accelerates the symmetric link establishment process. The bidirectional link setup under this new mechanism is significantly streamlined.
For persistently asymmetric links (e.g., unidirectional links), CLC-OLSR implements a state maintenance mechanism. Node A maintains a counter for its asymmetric neighbor B. This counter increments at the end of each HELLO interval. If the counter reaches a pre-defined threshold, node B is deemed an invalid, non-upgradable neighbor and is purged from the asymmetric list, thereby reducing the length of subsequent HELLO messages.
2. Route Discovery Mechanism Based on Cross-Layer Information Reuse
Building routes in standard OLSR relies on network-layer control message exchanges, which incur delay and overhead. CLC-OLSR proposes a cross-layer route discovery mechanism that leverages the topological information inherent in the node attachment process of the Two-Hop TDMA protocol.
Consider a two-hop node A joining the network. It sends a network entry request to the central node C via a one-hop node B. In the TDMA frame, the original source address (Initial_src) remains as A’s address, while the immediate source address (Src) is updated to B’s address by the forwarding node. When the central node C processes this request, it can directly infer the complete path A-B-C from these two addresses. This information is passed to the network layer, enabling C to autonomously construct a routing table entry for node A without relying on any network-layer HELLO messages.
Once the central node has built its routing table autonomously, it ceases broadcasting HELLO messages. To ensure other nodes can still maintain correct neighbor states for the central node, a one-hop node constructs an “equivalent HELLO message” on behalf of the central node. The source address is set to the central node’s address, and the symmetric neighbor list is populated with all one-hop node addresses, information readily available from the Beacon slot allocation table broadcast by the central node. This equivalent message is then processed normally by the network layer stack, maintaining compatibility with standard OLSR neighbor state maintenance.
3. Adaptive HELLO Interval Optimization Mechanism Based on Cross-Layer Triggering
Traditional adaptive interval schemes couple topology sensing with interval adjustment: the interval is adjusted based on neighbor list stability, but the neighbor list itself requires HELLO messages to build. This creates a dependency loop that limits optimization. Furthermore, network-layer sensing is slow and vulnerable to transient packet loss. CLC-OLSR decouples these functions by using the data link layer for real-time topology monitoring and triggering network-layer responses.
The data link layer continuously monitors the state of the asymmetric neighbor set. Changes are detected through:
- Beacon Monitoring: Nodes track Beacon messages from neighbors. Missed Beacons beyond a threshold indicate a neighbor has left.
- Network Entry Events: Hearing a new node’s entry request frame immediately adds that node to the listener’s asymmetric list.
- Mobility Detection: When a node re-associates with a different parent, the central node’s updated Beacon slot table reveals the change to the old neighbors, prompting them to update their lists.
Upon detecting a change, the data link layer immediately sends an event notification to the network layer. The network layer responds by instantly sending a HELLO message to propagate the new topology information rapidly, ensuring fast route convergence. When the topology is stable, the HELLO interval remains at a configured maximum, minimizing control overhead. A fault-tolerant mechanism runs at the data link layer with a 1-second cycle, using thresholds to avoid false triggers from occasional packet loss, thus balancing responsiveness and stability for China UAV drone networks.
Theoretical Performance Analysis
We analyze the control overhead of CLC-OLSR theoretically, comparing it with standard OLSR and the Low-Overhead OLSR for Small UAV (LS-OLSR) protocol. Let \( S_i \) denote the length of the HELLO message from node \( i \), \( T_i \) its sending interval, and \( N \) the total number of nodes sending HELLO messages.
For standard OLSR, the total control overhead \( L^{OLSR}_{Total} \) per unit time is:
$$L^{OLSR}_{Total} = \sum_{i=1}^{N^{OLSR}} \frac{S^{OLSR}_i}{T^{OLSR}}$$
where \( T^{OLSR} \) is a constant interval for all nodes.
LS-OLSR introduces an adaptive interval that increases when the network is stable, so the average interval \( \overline{T^{LS}} \) is larger than \( T^{OLSR} \). It does not change \( N \) or \( S_i \) significantly. Thus:
$$L^{LS}_{Total} = \sum_{i=1}^{N^{LS}} \frac{S^{LS}_i}{T^{LS}_i} \approx \sum_{i=1}^{N^{OLSR}} \frac{S^{OLSR}_i}{T^{LS}_i} < \sum_{i=1}^{N^{OLSR}} \frac{S^{OLSR}_i}{T^{OLSR}} = L^{OLSR}_{Total}$$
since \( \frac{1}{N^{LS}}\sum_{i=1}^{N^{LS}} T^{LS}_i > T^{OLSR} \).
CLC-OLSR builds upon this by:
- Reducing HELLO message length \( S^C_i < S^{LS}_i \) via the asymmetric link optimization mechanism.
- Reducing the number of HELLO-sending nodes \( N^C < N^{LS} \) as the central node stops sending HELLOs.
- Employing a cross-layer triggered adaptive interval where the stable-state interval is a fixed maximum (equal to LS-OLSR’s final threshold), not a gradually increasing one, leading to a larger effective average interval \( \overline{T^C} > \overline{T^{LS}} \).
Therefore, the total overhead for CLC-OLSR is:
$$L^{C}_{Total} = \sum_{i=1}^{N^{C}} \frac{S^{C}_i}{T^{C}_i} < \sum_{i=1}^{N^{LS}} \frac{S^{LS}_i}{T^{LS}_i} = L^{LS}_{Total}$$
Consequently, we have the relationship \( L^{C}_{Total} < L^{LS}_{Total} < L^{OLSR}_{Total} \), proving the effectiveness of CLC-OLSR in reducing control overhead.
Simulation Verification and Analysis
To validate the performance of CLC-OLSR in a China UAV drone network context, simulations were conducted using OPNET 14.5. The protocol was compared against standard OLSR and LS-OLSR. A two-hop clustered network of 32 nodes was modeled, with 4 constant bit rate (CBR) data flows. Key simulation parameters are summarized in the table below.
| Parameter | Value |
|---|---|
| Node Communication Range | 5 km |
| Number of Network Nodes | 32 |
| Number of Data Flows | 4 |
| Data Packet Size | 256 Bytes |
| Data Link Layer Protocol | Centralized Two-Hop TDMA |
| Simulation Time | 500 s |
1. Network Control Overhead
The simulation results for normalized control overhead are shown below. CLC-OLSR consistently demonstrates the lowest overhead among the three protocols. By eliminating the central node’s HELLO messages, shortening HELLO packet length, and maintaining an optimal transmission interval, CLC-OLSR reduces the control channel burden. Quantitative analysis shows that CLC-OLSR achieves approximately a 13.3% reduction in control overhead compared to the LS-OLSR protocol, making it highly efficient for bandwidth-constrained China UAV drone links.
2. Network Throughput
Network throughput, measured in kilobits per second (kbps), was evaluated. CLC-OLSR achieved the highest throughput. The reduction in control traffic directly frees up channel bandwidth for actual data transmission. The efficient and stable routing table maintained by CLC-OLSR also minimizes routing delays and packet drops due to stale routes. The throughput of CLC-OLSR was measured to be about 3.7% higher than that of LS-OLSR, demonstrating a tangible improvement in data delivery capacity for China UAV drone mission payloads.
3. Network Packet Loss Rate
The packet loss rate is a critical metric for reliability. CLC-OLSR exhibited the lowest packet loss rate. Its fast convergence capability, enabled by cross-layer triggers, allows rapid route establishment and update upon topology changes. The decoupled design ensures quick response to node mobility or link failures, which are common in dynamic China UAV drone swarms. The stable, long intervals during quiet periods do not harm responsiveness because changes are detected independently at the data link layer. Simulation data indicates that CLC-OLSR reduces the packet loss rate by roughly 11.3% compared to the LS-OLSR protocol.
Conclusion
This paper addressed the challenge of high control overhead in proactive routing for two-hop China UAV drone ad-hoc networks. We proposed the CLC-OLSR protocol, a cross-layer collaborative design that synergizes a Two-Hop TDMA data link layer with an enhanced OLSR network layer. The protocol incorporates three core mechanisms: an Asymmetric Link Optimization mechanism to simplify link setup and shorten messages, a Cross-Layer Information Reuse mechanism for autonomous route discovery at the central node, and a Cross-Layer Triggered Adaptive HELLO Interval mechanism to decouple topology sensing from message scheduling. Theoretical analysis confirms its superiority in reducing overhead. Simulation results validate that CLC-OLSR outperforms both standard OLSR and the LS-OLSR protocol, achieving significantly lower control overhead, higher network throughput, and lower packet loss rate. These improvements make CLC-OLSR a highly suitable routing solution for the specific two-hop clustered topology prevalent in many China UAV drone application scenarios. Future work will explore integrating physical-layer link quality metrics to further enhance routing decisions and network resilience for next-generation China UAV drone fleets.