An unmanned self-organizing network cross-layer routing method based on space domain risk perception

Abstract

The application discloses a kind of unmanned self-organizing network cross-layer routing method based on space risk perception, belong to wireless communication and flight self-organizing network technical field.Correlation data is acquired, and candidate directed link set is constructed;Based on limited snap estimation interference and noise covariance matrix, through diagonal loading robust MVDR beam forming, link space filtering signal-to-interference-and-noise ratio, effective transmission rate and single-hop delay are obtained;According to the direction vector of received end expected signal and interference, three-dimensional angle separation degree is calculated, and is mapped as angle domain coupling factor and space risk cost;Single-hop delay and space risk cost are fused as link weight, and risk perception weighted directed graph is established, and interference-resistant multihop routing is generated using shortest path algorithm.The application introduces physical layer interference resistance, angle domain separability into routing decision, avoids high-risk link of expected signal and interference near collinearity, considers transmission efficiency, improves path angle domain safety margin, link stability and end-to-end transmission performance.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication and flight ad hoc network technology, and in particular relates to a cross-layer anti-interference routing method under malicious interference environment, especially a cross-layer routing method for unmanned ad hoc networks based on airspace risk perception. Background Technology

[0002] In recent years, drone swarms and ad hoc networks have been widely used in emergency communication, collaborative detection, low-altitude monitoring, disaster relief, and integrated air-space-ground communication. Ad hoc networks typically consist of multiple drone nodes with wireless transceiver and relay capabilities, with each node transmitting information via multi-hop communication. However, in complex electromagnetic warfare environments, malicious jamming nodes can reduce the received signal-to-interference-plus-noise ratio (SNR) of drone nodes by emitting high-power co-channel or adjacent-channel jamming signals, causing a decrease in link transmission rate, packet loss, or even communication interruption. Because the position and attitude of drone nodes are constantly changing, the incident directions of the desired and jamming signals at the receiving nodes also change, making the network link's anti-jamming capability exhibit significant spatial correlation and time-varying characteristics.

[0003] Existing UAV communication systems typically employ anti-interference techniques at the physical layer, such as adaptive beamforming, interference null formation, power control, frequency switching, and channel selection, to suppress interference. Traditional physical layer beamforming methods generally achieve good interference suppression when the desired signal and the interfering signal have sufficient angular spacing. However, when the direction of the desired signal and the direction of the interference are nearly coincident or approximately collinear from the receiver node's perspective, strong coupling easily occurs between the main lobe preservation constraint and the interference suppression constraint. In this case, even with robust adaptive beamforming methods, problems such as insufficient interference null depth, decreased desired signal gain, and deteriorated output signal-to-interference-plus-noise ratio (SNR) may occur. Changes in node position, movement of interfering nodes, errors in angle of arrival estimation, and lag in channel state information can further exacerbate these performance degradations.

[0004] At the network layer, existing routing methods for ad hoc networks typically use metrics such as inter-node distance, hop count, link transmission rate, remaining energy, link lifetime, or transmission delay for path selection. Some cross-layer routing methods introduce the physical layer's output signal-to-interference-plus-noise ratio (SIR) or link rate into the network layer to avoid links with poor current communication performance. However, instantaneous SIR, rate, or delay primarily reflect the transmission capability of candidate links in their current state and cannot directly characterize the potential spatial vulnerability when the desired signal direction is nearly collinear with the interference direction. Therefore, even if a candidate link currently has a low transmission delay, it may rapidly degrade due to insufficient angular safety margin after slight node displacement, changes in interference direction, or deviations in angle of arrival estimation.

[0005] Existing network layer routing costs typically require accumulation and comparison across multi-hop paths. However, the signal direction, interference direction, and angular relationships obtained at the physical layer are spatial geometric information, which cannot be directly used as edge weights for path optimization at the network layer. A key challenge in achieving coordinated physical layer spatial anti-interference processing and network layer routing decision-making is how to convert the three-dimensional angular domain separability of the desired signal and interference signal at the receiving node into a non-negative, accumulative, and optimizable link cost, and how to uniformly characterize this link cost along with the transmission delay or link rate after beamforming. In summary, existing technologies still suffer from shortcomings such as the disconnect between physical layer spatial filtering and network layer routing selection, the lack of explicit quantification of angular domain risks in candidate links, and the difficulty in actively avoiding high-risk, near-collinear links. These limitations make it difficult to simultaneously consider the current transmission efficiency, spatial anti-interference margin, and end-to-end routing stability of flight ad hoc networks under strong directional interference environments. Summary of the Invention

[0006] The purpose of this invention is to solve the problems mentioned in the background art and to propose a cross-layer routing method for unmanned self-organizing networks based on airspace risk perception.

[0007] To achieve the objective of this invention, a cross-layer routing method for unmanned self-organizing networks based on airspace risk awareness is disclosed, comprising the following steps:

[0008] Step 1: Obtain the three-dimensional position information, candidate link distance, expected signal arrival direction, interference signal arrival direction, channel state information, and array received snapshot data of each UAV node and interference node in the flight ad hoc network, and construct a candidate directed link set based on the physical reachability relationship between nodes;

[0009] Step 2: For each candidate link in the candidate directed link set, estimate the interference plus noise covariance matrix using array received snapshot data, diagonally load the interference plus noise covariance matrix, and use the robust minimum variance distortionless response beamforming method to calculate the output signal-to-interference-plus-noise ratio of the candidate link after spatial filtering. Determine the effective transmission rate and single-hop transmission delay based on the output signal-to-interference-plus-noise ratio.

[0010] Step 3: From the perspective of the receiving node of each candidate link, construct the desired signal direction vector and the interference direction vector based on the three-dimensional position information, calculate the three-dimensional angular separation degree between the desired signal direction vector and the interference direction vector, and map the three-dimensional angular separation degree into the angular domain coupling factor and the non-negative spatial domain risk cost.

[0011] Step 4: Combine the single-hop transmission delay of each candidate link with the airspace risk cost to form the cross-layer link weight, establish a risk-aware weighted directed graph, and perform a weighted shortest path search on the risk-aware weighted directed graph to obtain an anti-interference multi-hop route from the source node to the destination node.

[0012] Furthermore, step 1 specifically includes:

[0013] exist At that moment, the first The three-dimensional position of each drone node is represented as follows: The three-dimensional position of the interfering node is represented as ;

[0014] Candidate link distances are determined based on the Euclidean distance between any two UAV nodes, and ordered node pairs that satisfy the preset maximum communication distance and basic connectivity constraints are selected as candidate directed links.

[0015] The drone nodes in the flight ad hoc network are all configured with A uniform linear array consisting of n elements, with an element spacing of n. ;in The carrier wavelength.

[0016] Furthermore, for the sending node Point to the receiving node Candidate links, receiving nodes Observed sending node The azimuth and elevation angles are respectively and The steering vector of a uniform linear array is:

[0017] (1)

[0018] in, The direction cosine of the signal incident direction along the array axis is represented by the azimuth angle. and pitch angle Uniquely certain;

[0019] A line-of-sight-dominated UAV air link is constructed using the Friis free-space model; nodes To the node The free space path gain is ,in For nodes To the node The distance, and the corresponding valid channel vector, are represented as:

[0020] (2)

[0021] Interference source to receiving node The interference channel vector is

[0022] (3)

[0023] in, Interference source To the node distance, and For receiving nodes Observed sources of interference The azimuth and elevation angles.

[0024] Furthermore, the robust minimum variance distortionless response beamforming method in step 2 specifically includes:

[0025] Quick snapshot based on interference and noise

[0026] (4)

[0027] in, Indicates interference power. Indicates interference signal. It is additive white Gaussian noise;

[0028] Estimate the interference plus noise covariance matrix:

[0029] (5)

[0030] in, The conjugate transpose is a standard operation for array signals;

[0031] The robust covariance matrix after diagonal loading is ,in , The diagonal loading coefficient is used. It is the identity matrix;

[0032] The desired direction steering vector, after taking into account the angle of arrival estimation error, is denoted as... Then, by solving the linear equations Find the intermediate vector and in accordance with Obtain the robust minimum variance distortion-free response beam weight vector.

[0033] Furthermore, candidate links The output signal-to-interference-plus-noise ratio after robust minimum variance distortionless response beamforming is:

[0034] (6)

[0035] in, Represents a node The transmission power, This represents the squared array gain of the target channel. This refers to the interference power that leaks into the receiver after beam weighting. For noise power, Let L be the squared L2 norm of the MVDR weight vector;

[0036] The corresponding ideal spectral efficiency is

[0037] (7)

[0038] After considering the modulation and coding upper limit and protocol overhead, the effective link rate is:

[0039] (8)

[0040] in, For system bandwidth, This is the link efficiency factor. For maximum spectral efficiency; if the link rate is below the minimum available rate threshold. If the link is unavailable, then the effective transmission rate is:

[0041] (9)

[0042] The proposed routing scheme considers both transmission efficiency and corner domain robustness when constructing link costs, rather than simply selecting the geometric shortest path or the path with the minimum instantaneous delay; for each physically reachable directed link... The actual single-hop communication delay consists of propagation delay, processing delay, and packet transmission delay:

[0043] (10)

[0044] in, Link distance, For the speed of transmission, To handle latency, For group length, This represents the effective transmission rate based on MVDR.

[0045] Furthermore, in step 3, for the node sending the message... Point to the receiving node For the candidate links, the desired signal direction vector and interference direction vector are as follows:

[0046] , (11)

[0047] in, For the location of the sending node, For the location of the receiving node, Location of the jammer;

[0048] Therefore, the three-dimensional angular separation between the desired signal direction and the interference direction is calculated as follows:

[0049] (12)

[0050] Among them, superscript This is a vector transpose operation. For vector 2-norm, This is a stable bias value used to avoid a denominator of zero; The smaller the value, the higher the degree of directional overlap, making it easier to damage the main lobe of the desired signal during spatial interference suppression.

[0051] Furthermore, the 3D angular separation is mapped to a Gaussian angular domain coupling factor:

[0052] (13)

[0053] in, The antenna's half-power beamwidth; when the desired signal and the jammer are approximately in the same direction, the coupling factor approaches 1; as the angular separation increases, the coupling factor decreases rapidly;

[0054] Based on this, in order to encourage nodes to actively reject such links when selecting routes, the spatial domain risk cost is reconstructed as follows:

[0055] (14)

[0056] in, The weighting adjustment coefficient for the risk term is used to control the contribution of corner domain risk to the overall routing metric; this form imposes a sharply increased penalty on links close to the collinear failure boundary, while adding only a small cost to links with sufficient corner separation.

[0057] Furthermore, in step 4, candidate links The cross-layer link weights are:

[0058] (15)

[0059] For candidate links that are unavailable or physically unreachable, the weight of the cross-layer link is set to infinity; for the remaining candidate links, a risk-aware weighted directed graph is constructed with the drone node as the vertex, the candidate directed links as the directed edges, and the weight of the cross-layer link as the edge weight.

[0060] Furthermore, a Dijkstra-type weighted shortest path algorithm is adopted. For the set of feasible paths, from the source node to the destination node Find the optimal path from the set of feasible paths that satisfies the following formula. :

[0061] (16)

[0062] And determine the bottleneck throughput of the path based on the optimal path. Actual end-to-end delay and path minimum angular separation ,in:

[0063] (17)

[0064] (18)

[0065] (19)

[0066] in, Indicate candidate link Cross-layer link weights, Indicates the effective transmission rate. This represents the actual single-hop communication delay. This indicates the 3D angular separation.

[0067] Furthermore, after obtaining the anti-interference multi-hop route, the anti-interference multi-hop route is written into or updated to the routing table or forwarding table of the flight ad hoc network; when the UAV node location, interference node location, angle of arrival information, channel state information or array received snapshot data are updated, steps 1 to 4 are re-executed to update the anti-interference multi-hop route.

[0068] Compared with existing technologies, the significant advancements of this invention are as follows: 1) This invention transforms the three-dimensional angular domain separability between the desired signal direction and the interference direction from the receiving node's perspective into a non-negative spatial risk cost, enabling physical layer spatial anti-interference information to directly participate in network layer path calculation, achieving cross-layer coordination between physical layer beamforming and network layer routing; 2) This invention employs a robust MVDR beamforming method considering finite snapshot covariance matrix estimation, angle of arrival estimation error, and diagonal loading to evaluate the signal-to-interference-plus-noise ratio, effective transmission rate, and transmission delay of candidate links, improving the reliability of link performance evaluation under non-ideal array receiving conditions; 3) This invention uses a risk cost that asymptotically increases as the desired signal direction and the interference direction tend to be collinear, enabling the identification of instantaneous... 4) This invention uses the link transmission delay and spatial risk cost together to form a non-negative cross-layer edge weight, which can be directly solved using Dijkstra's shortest path algorithm. This reduces the sensitivity of routing selection to node movement, interference node movement, angle of arrival estimation error and channel state information aging; 5) In the case of high UAV node density and many selectable relay nodes, this invention can transform network topology redundancy into spatial bypass capability, avoid near-collinear bottleneck links that are prone to degradation in robust MVDR beamforming, and thus take into account both path robustness and end-to-end bottleneck throughput capability.

[0069] To more clearly illustrate the functional characteristics and structural parameters of the present invention, further explanation is provided below in conjunction with the accompanying drawings and specific embodiments. Attached Figure Description

[0070] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0071] Figure 1 This is a schematic diagram of a three-dimensional system model of a self-organizing network in flight under malicious interference environment according to the present invention;

[0072] Figure 2 This is a schematic diagram of the overall architecture and processing flow of the present invention;

[0073] Figure 3 These are the bottleneck throughput curves of the present invention and the comparative method under different interference locations;

[0074] Figure 4 This is the bottleneck throughput curve of the present invention and the comparative method under varying node density.

[0075] Figure 5 This is the cumulative distribution function curve of the present invention and the comparative method. Detailed Implementation

[0076] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0077] This invention proposes a cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness. It evaluates the anti-interference performance of each candidate link under non-ideal array reception conditions using robust adaptive beamforming, calculates the three-dimensional angular separation between the desired signal direction and the interference direction from the receiving node's perspective, and maps it to a non-negative spatial risk cost that can be accumulated and optimized at the network layer. This invention uses single-hop transmission delay and spatial risk cost together to constitute the cross-layer link weights. While considering current transmission efficiency, it reserves angular safety margins for physical layer spatial filtering, actively avoiding high-risk links where the desired signal direction and the interference direction are nearly collinear. Specific implementation steps are also provided.

[0078] Step 1: Construct a flight self-organizing network model and obtain candidate link parameters.

[0079] Step 2: Evaluate the anti-interference performance of candidate links using robust MVDR beamforming.

[0080] Step 3: Calculate the three-dimensional angular domain separability of candidate links and construct the spatial domain risk cost.

[0081] Step 4: Construct cross-layer link weights and obtain interference-resistant multi-hop routes.

[0082] Step 1: Construct a flight self-organizing network model and obtain candidate link parameters.

[0083] Without loss of generality, consider a multi-hop communication network for unmanned aerial vehicles (UAVs) deployed in a wide-area three-dimensional airspace, with a total number of nodes. The set of nodes is There exists a highly maneuverable, high-power co-frequency, co-directional interference source in the airspace of all nodes, denoted as . .

[0084] exist At any time, any node Position in a three-dimensional Cartesian coordinate system is defined as In the formula, and These represent the horizontal and vertical coordinates of the drone's location on the horizontal plane, respectively. This indicates the absolute altitude of the drone.

[0085] To meet the constraints of lightweight drone platforms, it is now assumed that all drone segments within the cluster are equipped with... Uniform linear array of array elements, array The array elements are arranged along the axial direction, with an element spacing of [missing information]. .in This is the carrier wavelength. For the receiving end... Observed transmitter Its azimuth and elevation angles are denoted as follows: Under far-field narrow-band conditions, the ULA steering vector can be written as:

[0086] (1)

[0087] in, The direction cosine of the signal incident direction along the array axis is represented by the azimuth angle. and pitch angle The only certainty.

[0088] A line-of-sight-dominated UAV air link is constructed using the Friis free-space model. Node To the node The free space path gain is The corresponding legal channel vector is represented as

[0089] (2)

[0090] Similarly, interference source to receiving node The interference channel vector is

[0091] (3)

[0092] For a candidate link Receive node The array sampling signal can be written as

[0093] (4)

[0094] in, For nodes The transmission power, For the power of the interference source, To legally transmit signals, This is an interference signal. It is Gaussian white noise.

[0095] Given a certain receiving strategy ,link The output signal-to-interference-plus-noise ratio is denoted as The corresponding ideal spectral efficiency is After considering the modulation and coding limits and protocol overhead, the effective link rate is: .in, For system bandwidth, This is the link efficiency factor. To achieve maximum spectral efficiency. If the link rate is below the minimum available rate threshold. If the link is not found, it will be considered unavailable.

[0096] (5)

[0097] Link latency consists of propagation latency, node processing latency, and packet transmission latency: .in, At the speed of light, For node processing and protocol stack forwarding latency, This is the length of the data packet. If... ,but .

[0098] For a path from the source node to the destination node Its end-to-end bottleneck throughput is defined as The end-to-end communication latency of the path is .

[0099] Step 2: Evaluate the anti-interference performance of candidate links using robust MVDR beamforming.

[0100] For links If the receiver directly uses a traditional single-antenna receiver, the power of the interference source will directly enter the receiver, causing the link to degrade rapidly under strong interference. To improve spatial anti-interference capability, this invention employs robust MVDR beamforming, which minimizes the output power of interference plus noise while maintaining distortion-free reception of the signal in the desired direction. The ideal MVDR problem can be expressed as... .in, For the beam weight vector, The interference plus noise covariance matrix, The direction guide vector for the desired signal direction.

[0101] In practical UAV networks, direction-of-arrival (DOA) estimation errors, a limited number of snapshots, and strong interference can all lead to unstable covariance matrix estimation. Therefore, this invention introduces DOA perturbation, finite snapshot covariance estimation, and diagonal loading. The desired direction estimate is... ,in Interference plus noise snapshot is The covariance matrix is ​​estimated as follows: The robust covariance matrix after diagonal loading is: ,in , The diagonal loading coefficients are then used. This is followed by a linear equation. Find the intermediate vector The MVDR weight vector is obtained by normalization. .

[0102] MVDR output SINR is

[0103] (6)

[0104] Then according to the formula Obtain the effective link rate.

[0105] Step 3: Calculate the three-dimensional angular domain separability of candidate links and construct the spatial domain risk cost.

[0106] Although robust MVDR can create spatial nulls in the direction of interference, its suppression capability is not solely determined by the interference power, but also strongly depends on the angular separability of the desired signal direction and the interference direction from the receiver array's perspective. When the two approximately coincide in three-dimensional space, the array needs to simultaneously maintain the desired signal gain and suppress the interference energy in the adjacent directions. This leads to the coupling of the main lobe preservation constraint and the null formation constraint, resulting in significant spatial vulnerability of the link. In this case, using only instantaneous SINR or transmission delay as routing weights is insufficient to promptly reflect the potential physical layer risks of the link approaching the collinearity failure boundary.

[0107] To characterize this airspace vulnerability, this invention considers starting from the launch node. to receiving node The link, in which , and These represent the positions of the transmitter, receiver, and jammer, respectively. The desired signal direction vector and the jamming direction vector observed from the receiver are defined as follows: , Therefore, the three-dimensional angular separation between the desired signal direction and the interference direction is calculated as follows:

[0108] (7)

[0109] in, This is a stable bias value used to avoid a denominator of zero. The smaller the value, the higher the degree of directional overlap, making it easier to damage the main lobe of the desired signal during spatial interference suppression. To map the angular separation degree to a cost term available to the network layer, a Gaussian angular domain coupling factor is introduced:

[0110] (8)

[0111] in, Let be the antenna's half-power beamwidth. When the desired signal and the jammer are approximately in the same direction, the coupling factor approaches 1; as the angular separation increases, the coupling factor decreases rapidly. Based on this, to encourage nodes to actively exclude such links during routing, the spatial domain risk cost is reconstructed as:

[0112] (9)

[0113] in, This is used to control the contribution of corner domain risk to the overall routing metric. This form imposes a sharply increased penalty on links near the collinear failure boundary, while adding only a small cost to links with sufficient corner separation. Therefore, the proposed risk cost can guide the routing algorithm to proactively avoid links that are particularly vulnerable to spatial domain interference.

[0114] Step 4: Construct cross-layer link weights and obtain interference-resistant multi-hop routes.

[0115] After obtaining the single-hop robust MVDR performance and corner-domain risk metrics, the network layer merges them into a unified link cost for routing. On one hand, the effective link rate determines the packet transmission delay, reflecting the link's instantaneous communication capability under current channel, interference, and receive filtering conditions. On the other hand, the corner-domain risk cost characterizes the link's sensitivity to node movement, DOA estimation errors, and CSI aging, thus indicating its potential failure probability. These two factors correspond to current transmission efficiency and future robustness margin, respectively, and therefore should be jointly characterized within a unified cross-layer cost framework.

[0116] Therefore, the proposed routing scheme considers both transmission efficiency and corner domain robustness when constructing link costs, rather than simply selecting the geometric shortest path or the path with the minimum instantaneous delay. Subsequently, a Dijkstra-type shortest path search algorithm is used on the weighted directed graph to obtain the minimum-cost route. For each physically reachable directed link... The actual single-hop communication delay consists of propagation delay, processing delay, and packet transmission delay:

[0117] (10)

[0118] in, Link distance, For the speed of transmission, To handle latency, For group length, This represents the effective transmission rate based on MVDR. To simultaneously consider communication latency and vulnerability to spatial domain interference, the link... The weights are defined as follows:

[0119] (11)

[0120] in, This represents the corner domain risk cost obtained in the previous step. For unreachable links, it will... Set it to infinity so that it cannot be selected during the routing process.

[0121] After assigning risk-aware weights to all feasible links, starting from the source node to the destination node Perform a Dijkstra-type weighted shortest path search. The resulting optimal path is represented as:

[0122] (12)

[0123] The bottleneck throughput, actual end-to-end latency, and minimum corner separation of the selected path are as follows:

[0124] (13)

[0125] (14)

[0126] (15)

[0127] Example

[0128] The specific embodiments of the present invention are as follows. The simulation uses Python software, and the parameter settings do not affect the generality. Considering a... In the simulated three-dimensional free space region, there exists a high-power suppression interference source collinear with both the transmitter and receiver. The default power is Assuming each drone is equipped with a four-element ULA array, and the drone node transmit power is at... Fluctuations within a range, the system operates at Carrier frequency, channel bandwidth The positions of the transmitter and receiver are fixed. Noise power is determined by thermal noise spectral density, system bandwidth, and... The noise figure is calculated. Robust MVDR is used. The covariance matrix is ​​estimated using a snapshot with added interference and noise. The standard deviation of the DOA estimation error is set as follows: The diagonal loading coefficient is The service packet length is The single-hop processing latency is The maximum single-hop communication distance is Each set of parameters is executed. A second Monte Carlo experiment was conducted to ensure the stability of the statistical results.

[0129] This invention evaluates the performance of various schemes under the following three simulation scenarios. First, with a fixed total number of UAV nodes, the spatial coordinates of the jammer are scanned to study the impact of the jammer's location on end-to-end throughput and transmission latency. Second, the jammer is positioned near the extension of the source-destination axis, and the number of UAV nodes is varied to verify whether increasing relay density can effectively translate into stronger route detour capabilities. Third, with a fixed number of nodes and jammer location, the jammer's transmit power is scanned over a wide range to characterize the survival threshold of different strategies as interference intensity continuously increases.

[0130] Simulations compare the following five schemes:

[0131] SISO_D: This scheme uses a single antenna for reception and uses time delay as the link weight. It lacks spatial domain interference immunity and is considered the lowest complexity baseline scheme.

[0132] TxDiversity_D: Inspired by diversity methods, it employs a simplified diversity gain model to improve the desired signal power. Similar to SISO_D, the receiver does not implement spatial nulling, and the link weight remains the time delay.

[0133] MVDR_G: The physical layer uses robust MVDR, and the route weight is determined solely by geometric distance.

[0134] MVDR_D: The physical layer uses robust MVDR, and the route weight is set to the link delay after MVDR processing.

[0135] MVDR_A: The physical layer adopts robust MVDR, and the routing weight is composed of the sum of the link delay after MVDR processing and the corner domain risk cost. This scheme is the method proposed in this invention.

[0136] Figure 1 This is a schematic diagram of a three-dimensional system model of a self-organizing flight network under malicious interference conditions, based on the present invention. The scene includes interference nodes. It communicates with multiple UAV nodes, each distributed in a three-dimensional airspace, and their spatial locations are determined by... axis, shaft and The coordinate system is represented by axes. The drone icon represents a node in a flight ad hoc network. The gray dashed lines between nodes represent candidate communication links formed under constraints such as communication distance. The blue solid lines represent the desired multi-hop communication links selected by the spatial domain risk-aware cross-layer routing method according to this invention. The red lines represent interfering nodes. Interference links pointing to the corresponding receiving node. (UAV receiving node) For example, sending node To the receiving node To form a desired communication link, its link distance is denoted as... Interference nodes To the receiving node An interference link is formed, and its distance is denoted as . Receive node According to the sending node and interference nodes The three-dimensional spatial position is used to determine the direction of the desired signal and the direction of the interference signal, and the three-dimensional angular separation between them is calculated.

[0137] Figure 2 This is a schematic diagram of the overall architecture and processing flow of the present invention. First, network status information and array parameters of the FANET are collected, including the location of UAV nodes, candidate link distances, DOA information, channel status information, and array received snapshot data. Then, a physical layer robust link evaluation is performed on each candidate link. Through covariance matrix estimation, diagonal loading, and robust adaptive beamforming, link performance indicators such as signal-to-interference-plus-noise ratio (SINR), effective transmission rate, and single-hop transmission delay after interference suppression are obtained. Simultaneously with the link performance evaluation, a spatial domain risk assessment is performed on the candidate links based on the three-dimensional spatial positional relationship between the transmitting node, receiving node, and interfering node. The three-dimensional angle between the desired signal direction and the interference direction from the receiving node's perspective is calculated, and the angular domain coupling factor and spatial domain risk cost are further determined. Next, the transmission delay and spatial domain risk cost of the candidate links are combined to form cross-layer link edge weights, establishing a risk-aware weighted directed graph. Finally, minimum cost path solving is performed on the weighted directed graph. Based on the solution results, a multi-hop communication path from the source node to the destination node is determined, and the corresponding routing table or forwarding table is updated. This process enables the physical layer's output link anti-interference capability and the spatial separability between signals and interference to participate in network layer routing decisions, thereby proactively avoiding high-risk links where the desired signal direction and interference direction are close to collinear while taking into account the current link transmission efficiency.

[0138] Figure 3The bottleneck throughput of different methods is presented when the jammer moves along the y-axis. The results show that the end-to-end throughput is affected not only by the received interference power but also significantly depends on the angular relationship between the desired signal, the interference signal, and the relay link selected by the routing strategy. In the SISO_D scheme, due to the lack of anti-interference capability of the single-antenna receiver, its bottleneck throughput is the lowest, only about 0.05–0.12 Mbps throughout the simulation. TxDiversity_D improves the desired signal strength through transmit diversity, thus outperforming SISO_D, but it cannot actively suppress interference from the jammer's direction; its bottleneck throughput is about 0.33–0.43 Mbps, still significantly lower than the MVDR-based schemes. Among the various MVDR schemes, MVDR_G shows the most significant throughput decrease because the geometric shortest path criterion tends to select relay links closer to the source and destination connection. When the jammer approaches such links, the distortion-free response constraint of MVDR is strongly coupled with the interference null constraint, making it difficult for the array to simultaneously maintain high desired signal gain and deep interference suppression. Figure 3 As can be seen, the near-collinearity condition leads to a significant decrease in throughput near y ≈ 19 km and y ≈ 31 km. MVDR_D uses the link delay processed by MVDR as the routing weight, thus alleviating the above problem. The MVDR_A proposed in this invention achieves the most stable and consistently high performance. Its throughput remains above approximately 1.95 Mbps throughout the entire jammer's trajectory; compared to MVDR_G and MVDR_D, performance degradation near vulnerable areas is significantly suppressed. This demonstrates that corner-domain risk measurement can help the routing layer identify and avoid links that easily evolve into throughput bottlenecks under near-collinear geometric relationships between signal and interference. By selecting relay links with more favorable spatial separation conditions, MVDR_A provides better conditions for MVDR null formation and maintains a higher end-to-end bottleneck rate when the jammer's location is uncertain.

[0139] Figure 4The throughput gain from node encryption is demonstrated to largely depend on whether the routing metric can effectively utilize the added spatial degrees of freedom under malicious interference conditions. With an increase in the number of drones, both SISO_D and TxDiversity_D achieve some degree of throughput improvement. However, their performance remains fundamentally limited because neither can implement spatial interference suppression at the receiver. MVDR_G exhibits a different trend: its bottleneck throughput decreases with increasing drone density. This is because as the number of candidate relay nodes increases, MVDR_G is more likely to select links closer to the source-destination connection. The MVDR beamformer struggles to maintain a distortion-free response in the desired direction and form effective interference nulls simultaneously, leading to a decrease in SINR after beamforming and a reduction in bottleneck throughput. Therefore, when the routing metric ignores corner domain risk, newly added candidate nodes may transform from "beneficial resources" into "harmful choices." In contrast, MVDR_D and MVDR_A both significantly benefit from node encryption because they select paths based on the transmission delay after MVDR processing. The proposed MVDR_A further incorporates corner domain risk costs into the routing decision, thereby continuing to improve bottleneck throughput. Compared to MVDR_D, MVDR_A can identify links with acceptable instantaneous latency but remain vulnerable due to insufficient angular spacing between the desired and interfering signals. By penalizing such concealed high-risk links, MVDR_A translates increased relay density into more reliable spatial rerouting capabilities.

[0140] Figure 5 The CDF curves illustrate the impact of angular domain coupling on route robustness. Specifically, the CDF curve for MVDR_A shifts significantly to the right towards larger angular intervals compared to MVDR_D, indicating that the proposed risk perception metric alters the statistical distribution of the selected paths. In particular, its left-tail probability is reduced, meaning that routes with minimal path minimum angular intervals occur less frequently under MVDR_A. Since the minimum path angular interval determines the most vulnerable spatial location in the entire route, this improved distribution suggests that MVDR_A achieves a more conservative and robust route selection from an angular domain perspective.

[0141] In summary, this invention proposes a cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness. Specifically, on each candidate link in the flight ad hoc network, robust adaptive beamforming is used to obtain the link performance after interference resistance. Simultaneously, the three-dimensional angular separability between the desired signal direction and the interference direction is calculated from the perspective of the receiving node, mapped to a spatial domain risk cost, and then combined with link delay to constitute network layer edge weights. Based on this, multi-hop routes resistant to strong directional interference are obtained. This invention demonstrates that in malicious interference environments, spatial domain security and throughput performance are not necessarily mutually exclusive; reasonable angular domain risk modeling can simultaneously enhance link stability and path transmission capability. The model of this invention is complete, the design method is reasonable and effective, and it can enhance link stability and path transmission capability under malicious interference. Specific implementation steps are also provided.

[0142] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0143] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A cross-layer routing method for unmanned ad hoc networks based on airspace risk awareness, characterized in that, Includes the following steps: Step 1: Obtain the three-dimensional position information, candidate link distance, expected signal arrival direction, interference signal arrival direction, channel state information, and array received snapshot data of each UAV node and interference node in the flight ad hoc network, and construct a candidate directed link set based on the physical reachability relationship between nodes; Step 2: For each candidate link in the candidate directed link set, estimate the interference plus noise covariance matrix using array received snapshot data, diagonally load the interference plus noise covariance matrix, and use the robust minimum variance distortionless response beamforming method to calculate the output signal-to-interference-plus-noise ratio of the candidate link after spatial filtering. Determine the effective transmission rate and single-hop transmission delay based on the output signal-to-interference-plus-noise ratio. Step 3: From the perspective of the receiving node of each candidate link, construct the desired signal direction vector and the interference direction vector based on the three-dimensional position information, calculate the three-dimensional angular separation degree between the desired signal direction vector and the interference direction vector, and map the three-dimensional angular separation degree into the angular domain coupling factor and the non-negative spatial domain risk cost. Step 4: Combine the single-hop transmission delay of each candidate link with the airspace risk cost to form the cross-layer link weight, establish a risk-aware weighted directed graph, and perform a weighted shortest path search on the risk-aware weighted directed graph to obtain an anti-interference multi-hop route from the source node to the destination node.

2. The cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness according to claim 1, characterized in that, Step 1 specifically includes: exist At that moment, the first The three-dimensional position of each drone node is represented as follows: The three-dimensional position of the interfering node is represented as ; Candidate link distances are determined based on the Euclidean distance between any two UAV nodes, and ordered node pairs that satisfy the preset maximum communication distance and basic connectivity constraints are selected as candidate directed links. The drone nodes in the flight ad hoc network are all configured with A uniform linear array consisting of n elements, with an element spacing of n. ;in The carrier wavelength.

3. The cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness according to claim 2, characterized in that, For the sending node Point to the receiving node Candidate links, receiving nodes Observed sending node The azimuth and elevation angles are respectively and The steering vector of a uniform linear array is: (1) in, The direction cosine of the signal incident direction along the array axis is represented by the azimuth angle. and pitch angle Uniquely certain; A line-of-sight-dominated UAV air link is constructed using the Friis free-space model; nodes To the node The free space path gain is ,in For nodes To the node The distance, and the corresponding valid channel vector, are represented as: (2) Interference source to receiving node The interference channel vector is (3) in, Interference source To the node distance, and For receiving nodes Observed sources of interference The azimuth and elevation angles.

4. The cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness according to claim 1, characterized in that, The robust minimum variance distortionless response beamforming method in step 2 specifically includes: Quick snapshot based on interference and noise (4) in, Indicates interference power. Indicates interference signal. It is additive white Gaussian noise; Estimate the interference plus noise covariance matrix: (5) in, The conjugate transpose is a standard operation for array signals; The robust covariance matrix after diagonal loading is ,in , The diagonal loading coefficient is used. It is the identity matrix; The desired direction steering vector, after taking into account the angle of arrival estimation error, is denoted as... Then, by solving the linear equations Find the intermediate vector and in accordance with Obtain the robust minimum variance distortion-free response beam weight vector.

5. A cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness according to claim 4, characterized in that, Candidate Links The output signal-to-interference-plus-noise ratio after robust minimum variance distortionless response beamforming is: (6) in, Represents a node The transmission power, This represents the squared array gain of the target channel. This refers to the interference power that leaks into the receiver after beam weighting. For noise power, Let L be the squared L2 norm of the MVDR weight vector; The corresponding ideal spectral efficiency is (7) After considering the modulation and coding upper limit and protocol overhead, the effective link rate is: (8) in, For system bandwidth, This is the link efficiency factor. For maximum spectral efficiency; if the link rate is below the minimum available rate threshold. If the link is unavailable, then the effective transmission rate is: (9) The proposed routing scheme considers both transmission efficiency and corner domain robustness when constructing link costs; for each physically reachable directed link... The actual single-hop communication delay consists of propagation delay, processing delay, and packet transmission delay: (10) in, Link distance, For the speed of transmission, To handle latency, For group length, This represents the effective transmission rate based on MVDR.

6. The cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness according to claim 1, characterized in that, In step 3, for the sending node Point to the receiving node For the candidate links, the desired signal direction vector and interference direction vector are as follows: ， (11) in, For the location of the sending node, For the location of the receiving node, Location of the jammer; Therefore, the three-dimensional angular separation between the desired signal direction and the interference direction is calculated as follows: (12) Among them, superscript This is a vector transpose operation. For vector 2-norm, This is a stable bias value used to avoid a denominator of zero; The smaller the value, the higher the degree of directional overlap, making it easier to damage the main lobe of the desired signal during spatial interference suppression.

7. A cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness according to claim 6, characterized in that, Mapping the 3D angular separation degree to a Gaussian angular domain coupling factor: (13) in, The antenna's half-power beamwidth; when the desired signal and the jammer are approximately in the same direction, the coupling factor approaches 1; as the angular separation increases, the coupling factor decreases rapidly; Based on this, in order to encourage nodes to actively reject such links when selecting routes, the spatial domain risk cost is reconstructed as follows: (14) in, This is the weighting adjustment coefficient for the risk item, used to control the contribution of corner domain risk to the overall routing metric.

8. A cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness according to claim 7, characterized in that, In step 4, candidate links The cross-layer link weights are: (15) For candidate links that are unavailable or physically unreachable, the weight of the cross-layer link is set to infinity; for the remaining candidate links, a risk-aware weighted directed graph is constructed with the drone node as the vertex, the candidate directed links as the directed edges, and the weight of the cross-layer link as the edge weight.

9. A cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness according to claim 8, characterized in that, A Dijkstra-type weighted shortest path algorithm is used. For the set of feasible paths, from the source node to the destination node Find the optimal path from the set of feasible paths that satisfies the following formula. : (16) And determine the bottleneck throughput of the path based on the optimal path. Actual end-to-end delay and path minimum angular separation ,in: (17) (18) (19) in, Indicate candidate link Cross-layer link weights, Indicates the effective transmission rate. This represents the actual single-hop communication delay. This indicates the 3D angular separation.

10. A cross-layer routing method for unmanned ad hoc networks based on spatial risk awareness according to any one of claims 1 to 9, characterized in that, After obtaining the anti-interference multi-hop route, write or update the anti-interference multi-hop route to the routing table or forwarding table of the flight ad hoc network; when the UAV node location, interference node location, angle of arrival information, channel state information or array received snapshot data are updated, repeat steps 1 to 4 to update the anti-interference multi-hop route.

Images