Multipath routing selection and load balancing control method and system for SRv6 network

Abstract

The present invention provides a multi-path routing and load balancing control method and system for SRv6 networks, the method comprising collecting link status information, calculating path comprehensive weights, selecting the optimal path for data transmission, and dynamically adjusting the path selection strategy according to the load conditions to achieve network load balancing, wherein the comprehensive weight is calculated by considering factors such as link delay, bandwidth utilization, and impedance, and the optimization model is used to solve the optimal path selection strategy. The system comprises a path monitoring module, a path selection module, a load balancing module, an optimization solution module, a strategy adjustment module, and a feedback optimization module, and realizes efficient multi-path routing and load balancing control through collaborative work. The present invention can effectively improve the performance and reliability of SRv6 networks.

Description

Technical Field

[0001] The present invention relates to the technical field of routing selection and load control, and in particular to a multipath routing selection and load balancing control method and system for an SRv6 network. Background Art

[0002] Segment Routing over IPv6 (SRv6) is an IPv6-based network architecture that introduces the concept of segment routing into IPv6 networks. SRv6 implements flexible and programmable traffic engineering by embedding "segment identifiers" (SIDs) in data packets.

[0003] SRv6 uses the scalability of IPv6 addresses to embed multiple segments, thereby achieving complex path control and service quality assurance. In SRv6 networks, multi-path routing and load balancing are achieved through the following technical foundations:

[0004] Segment Routing (SR): SR uses segment routing to define the forwarding path of data packets, avoiding the complexity of traditional label-based MPLS networks. SRv6 expands on this basis and uses IPv6 addresses as segment identifiers to achieve more flexible traffic engineering.

[0005] Segment Identifiers (SIDs): SIDs are the core of SRv6 and are used to specify the path that a data packet should follow. Each SID represents a network node or network function, and through these SIDs, the network can precisely control the traffic path.

[0006] Traffic engineering: SRv6 allows network administrators to define traffic paths and performance requirements based on demand, supporting end-to-end traffic engineering and service quality assurance.

[0007] Multipath routing: SRv6 supports multipath routing, which can find multiple paths in the network to improve network reliability and performance. Multipath routing usually selects the best path based on parameters such as bandwidth, delay, and load.

[0008] Load balancing: In multi-path routing, load balancing technology is used to evenly distribute traffic to multiple paths, thereby avoiding overload of a single path and improving the efficiency and stability of the overall network.

[0009] For multipath control of SRv6 network architecture, the existing technical routes mainly include:

[0010] 1. Policy-based multi-path routing selection: Existing technologies usually use static or policy-based multi-path routing selection methods. These methods select multiple paths based on preset rules or policies, which are suitable for specific traffic patterns and network conditions.

[0011] 2. Dynamic load balancing: The dynamic load balancing method in the prior art uses real-time traffic data and network status information to dynamically adjust the distribution of traffic, including traffic distribution based on traffic hashing, load sensing and other mechanisms.

[0012] 3. Routing selection based on optimization algorithms: Optimization algorithms (such as shortest path algorithm, bandwidth-delay optimization, etc.) are used to select the best path. These algorithms take into account various parameters of the network to achieve the optimal path selection.

[0013] 4. Segment Routing (SR) control plane: In an SRv6 network, the control plane usually uses IGP (Interior Gateway Protocol) or BGP (Border Gateway Protocol) to propagate routing information and SIDs to achieve path selection and load balancing.

[0014] Although SRv6 networks provide a flexible mechanism for multi-path routing and load balancing, existing technologies still face challenges in terms of dynamic adaptability, load balancing accuracy, computational complexity, control plane overhead, and path consistency. These challenges mainly include:

[0015] 1. Insufficient flexibility of static policies: Policy-based multi-path selection methods often lack flexibility and are difficult to adapt to dynamic changes in the network environment. Static policies may not be able to respond to traffic fluctuations and changes in network status in real time.

[0016] 2. Insufficient load balancing accuracy: Existing load balancing mechanisms may not accurately balance the traffic on each path, causing some paths to be overloaded while other paths are not fully utilized. This may reduce the overall performance and reliability of the network.

[0017] 3. Computational complexity of optimization algorithms: Although routing selection based on optimization algorithms can provide better performance, its computational complexity may be high, especially in large-scale networks, which may affect the efficiency of real-time routing decisions.

[0018] 4. Control plane overhead: In SRv6 networks, the control plane overhead may be large, especially in large-scale networks. The propagation of routing information and SIDs will bring additional overhead and affect the scalability of the network.

[0019] 5. Multi-path consistency issue: In actual deployment, it may be challenging to ensure the consistency and coordination of multi-path routing. Traffic on different paths may be inconsistent, thus affecting the quality of service.

[0020] To address these issues, new methods and technologies are urgently needed to improve the performance and efficiency of SRv6 networks. Summary of the invention

[0021] To achieve the purpose of the present invention, the present invention is implemented by the following technical solutions:

[0022] A multipath routing selection and load balancing control method for an SRv6 network, the method comprising the following steps:

[0023] S1. Collect path status information in the SRv6 network, where the path status information includes at least sub-link delay, sub-link bandwidth, and sub-link utilization;

[0024] S2. Calculate the comprehensive weight of each path based on the path status information. The comprehensive weight W p Calculated by the following formula:

[0025]

[0026] Among them, W p represents the comprehensive weight of path p; and They represent the weight coefficient, sublink delay, sublink bandwidth, sublink utilization and sublink impedance factor of the sublink numbered i under path p, respectively. p represents the total number of sub-links under path p; is the combined function of the sub-link states of the sub-link numbered i under the path p;

[0027] S3, select the optimal path for data transmission according to the comprehensive weight, and record the load of the current path;

[0028] S4. Based on the recorded load conditions, dynamically adjust the path selection strategy to balance the network load;

[0029] S5. Repeat steps S1 to S4 to dynamically monitor and execute multi-path routing selection and load balancing control.

[0030] Further, in step S2, the combination function of the sub-link status is calculated by the following formula:

[0031]

[0032] in, is the combined function of the sub-link states of the sub-link numbered i under path p, α p and β p is the adjustment parameter of path p, is the maximum sub-link delay of all sub-links under path p, is the adjustment parameter of the sub-link bandwidth utilization of the sub-link with sequence number i under path p;

[0033] Sublink Delay The values ​​are measured in real time.

[0034] Adjustment parameter α p Used to control the output range of the function and adjust the parameter β p Used to control sensitivity and adjust parameters Used to adjust the bandwidth utilization of the sub-link, adjusting parameter α p , adjustment parameter β p and adjustment parameters The value of is affected by real-time delay and real-time bandwidth, and is predetermined through fitting experiments.

[0035] Further, in step S2, the sub-link impedance factor is calculated according to the following formula:

[0036]

[0037] in, represents the sub-link impedance factor, λ p It is a parameter used to adjust the sensitivity of the algorithm, and its value is based on experience. is the average delay of all sub-links under path p, which is obtained by measuring the delay of all sub-links and then calculating the average value of the sub-link delay;

[0038] It is the adjustment parameter of the impedance of the sub-link with sequence number i under path p, which is used to adjust the sensitivity of the algorithm to the path impedance. The value is predetermined by fitting experiments.

[0039] Furthermore, the adjustment parameter α p , adjustment parameter β p , Adjustment parameters Tuning parameter λ p and adjustment parameters The initial value of is 1.

[0040] Further, in step S3, the optimal path is selected for data transmission according to the comprehensive weight value, including: solving an optimization model, the expression of the optimization model is:

[0041]

[0042] Among them, p *represents the multipath optimal load combination matrix; f(p) represents the multipath load delay intensity function; {P} represents the value range of all possible multipath load combination matrices; MIN[] represents the minimization solution operation, and MIN[f(p)] represents the multipath load combination matrix that makes the path load delay intensity function obtain the minimum value within the value range of all possible multipath load combination matrices, as the multipath optimal load combination matrix p * ; T p represents the transmission delay of path p, T max Indicates the maximum allowed delay, F p represents the load factor of path p.

[0043] Furthermore, the load factor is calculated by the following formula:

[0044]

[0045] Among them, F p represents the load factor of path p, m p is the total number of flows on path p; C j and D j They represent the flow rate and duration of the j-th flow respectively.

[0046] Furthermore, in step S5, the dynamic multi-path routing selection and load balancing control specifically includes:

[0047] S51, performing real-time load monitoring on each path and calculating the load change rate of the path;

[0048] S52, adjusting the flow transfer strategy based on the load change rate so that the load tends to be balanced;

[0049] S53: If it is detected that the load change rate of a certain path exceeds a threshold, a path with a lower load is preferentially selected for data transmission.

[0050] Further, in step S51, the load change rate is calculated by the following formula:

[0051]

[0052] Where ρ represents the load change rate, represents the initial load factor of path p, represents the current load factor of path p after a time interval Δt.

[0053] Further, step S52 specifically includes:

[0054] S521, calculate the load factors of all available paths, select the path with the largest load factor as the outbound target, select the path with the smallest load factor as the inbound target, and transfer the traffic;

[0055] S522, looping through step S521, gradually transferring the traffic on the path with a high load change rate to the target path until the load balancing condition is met.

[0056] The present invention also provides a multipath routing and load balancing control system for an SRv6 network, which is used to execute the multipath routing and load balancing control method for an SRv6 network, and the system includes:

[0057] Path monitoring module: used to monitor the link status information and load conditions of each path;

[0058] Path selection module: Based on the information provided by the monitoring module, it calculates the comprehensive weight of the path and selects the optimal path;

[0059] Load balancing module: used to dynamically adjust the path selection strategy to balance the load of each path;

[0060] Optimization solution module: based on the output of the path selection module, solve the optimal path selection strategy and generate a path routing table;

[0061] Strategy adjustment module: adjusts the path selection strategy in advance according to the load prediction results to avoid path congestion;

[0062] Feedback optimization module: monitors the effect of path selection in real time and continuously optimizes the path selection algorithm based on feedback information.

[0063] Beneficial Effects

[0064] The present invention provides a multipath routing selection and load balancing control method for SRv6 networks, which achieves significant technical effects through the following aspects:

[0065] 1. Improve network transmission efficiency: By collecting path status information in the SRv6 network, including sub-link delay, sub-link bandwidth, and sub-link utilization, the comprehensive weight of each path is accurately calculated to ensure that the data flow has the best transmission path in the network, reducing data transmission delays and bandwidth waste, thereby improving the transmission efficiency of the entire network.

[0066] 2. Realize dynamic load balancing: The present invention designs a dynamic adjustment mechanism based on the comprehensive weight of the path and the load factor. By real-time monitoring and adjusting the selection strategy of the network path, the load distribution of the network is made more uniform, avoiding the performance bottleneck caused by overloading of some paths, and improving the resource utilization and service quality of the network.

[0067] 3. Improve the robustness and adaptability of the network: Through real-time monitoring and dynamic adjustment of the path, the present invention can promptly respond to changes in the load in the network, quickly adjust the routing selection of the data flow, reduce the risk of network congestion caused by burst traffic, and enhance the network's adaptability to emergencies.

[0068] 4. Optimize network resource configuration: By building an optimization model and calculating the optimal multi-path load combination matrix, we ensure that the path with the smallest load delay is selected from all possible multi-path combinations, thereby optimizing the configuration of network resources to the maximum extent and improving the overall service quality of the network.

[0069] 5. Reduce the impact of path impedance: By calculating the path impedance factor and combining it with a comprehensive weight model, the present invention effectively reduces the negative impact of path impedance on route selection, thereby ensuring the stability and efficiency of data transmission, especially in complex network environments, where the advantages are more significant. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0071] In order to deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with examples. The examples are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention.

[0072] Embodiment 1

[0073] according to Figure 1 As shown, this embodiment provides a multipath routing selection and load balancing control method for an SRv6 network, the method comprising the following steps:

[0074] S1. Collect path status information in the SRv6 network, where the path status information includes at least sub-link delay, sub-link bandwidth, and sub-link utilization;

[0075] S2. Calculate the comprehensive weight of each path based on the path status information. The comprehensive weight W p Calculated by the following formula:

[0076]

[0077] Among them, W p represents the comprehensive weight of path p; and They represent the weight coefficient, sublink delay, sublink bandwidth, sublink utilization and sublink impedance factor of the sublink numbered i under path p, respectively. p represents the total number of sub-links under path p; is the combined function of the sub-link states of the sub-link numbered i under the path p;

[0078] S3, select the optimal path for data transmission according to the comprehensive weight, and record the load of the current path;

[0079] S4. Based on the recorded load conditions, dynamically adjust the path selection strategy to balance the network load;

[0080] S5. Repeat steps S1 to S4 to dynamically monitor and execute multi-path routing selection and load balancing control.

[0081] In step S2, the combination function of the sub-link status is calculated by the following formula:

[0082]

[0083] in, is the combined function of the sub-link states of the sub-link numbered i under path p, α p and β p is the adjustment parameter of path p, is the maximum sub-link delay of all sub-links under path p, is the adjustment parameter of the sub-link bandwidth utilization of the sub-link with sequence number i under path p;

[0084] Sublink Delay The values ​​are measured in real time.

[0085] Adjustment parameter α p Used to control the output range of the function and adjust the parameter β p Used to control sensitivity and adjust parameters Used to adjust the bandwidth utilization of the sub-link, adjusting parameter α p , adjustment parameter β p and adjustment parameters The value of is affected by real-time delay and real-time bandwidth, and is predetermined through fitting experiments.

[0086] In step S2, the sub-link impedance factor is calculated according to the following formula:

[0087]

[0088] in, represents the sub-link impedance factor, λ p It is a parameter used to adjust the sensitivity of the algorithm, and its value is based on experience. is the average delay of all sub-links under path p, which is obtained by measuring the delay of all sub-links and then calculating the average delay of the sub-links;

[0089] It is the adjustment parameter of the impedance of the sub-link with sequence number i under path p, which is used to adjust the sensitivity of the algorithm to the path impedance. The value is predetermined by fitting experiments.

[0090] Adjustment parameter α p , adjustment parameter β p , Adjustment parameters Tuning parameter λ p and adjustment parameters The initial value of is 1.

[0091] In step S3, the optimal path is selected for data transmission according to the comprehensive weight value, including: solving an optimization model, the expression of the optimization model is:

[0092]

[0093] Among them, p * represents the multipath optimal load combination matrix; f(p) represents the multipath load delay intensity function; {P} represents the value range of all possible multipath load combination matrices; MIN[] represents the minimization solution operation, and MIN[f(p)] represents the multipath load combination matrix that makes the path load delay intensity function obtain the minimum value within the value range of all possible multipath load combination matrices, as the multipath optimal load combination matrix p * ; T p represents the transmission delay of path p, T max Indicates the maximum allowed delay, F p represents the load factor of path p.

[0094] The load factor is calculated by the following formula:

[0095]

[0096] Among them, F p represents the load factor of path p, m p is the total number of flows on path p; C j and D j They represent the flow rate and duration of the j-th flow respectively.

[0097] In step S5, the dynamic multi-path routing selection and load balancing control specifically includes:

[0098] S51, performing real-time load monitoring on each path and calculating the load change rate of the path;

[0099] S52, adjusting the flow transfer strategy based on the load change rate so that the load tends to be balanced;

[0100] S53: If it is detected that the load change rate of a certain path exceeds a threshold, a path with a lower load is preferentially selected for data transmission.

[0101] In step S51, the load change rate is calculated by the following formula:

[0102]

[0103] Where ρ represents the load change rate, represents the initial load factor of path p, represents the current load factor of path p after a time interval Δt.

[0104] Step S52 specifically includes:

[0105] S521, calculate the load factors of all available paths, select the path with the largest load factor as the outbound target, select the path with the smallest load factor as the inbound target, and transfer the traffic;

[0106] S522, looping through step S521, gradually transferring the traffic on the path with a high load change rate to the target path until the load balancing condition is met. Specific embodiment 2

[0108] In an actual SRv6 network, network administrators want to optimize the data flow distribution between multiple paths to improve transmission efficiency and achieve load balancing. The network includes multiple sub-links, and the status information of each path (such as sub-link delay, sub-link bandwidth, and sub-link utilization) can be obtained in real time through network monitoring tools.

[0109] Step S1: Collect path status information;

[0110] The network administrator uses the monitoring tools deployed in the network to regularly collect the status information of each path. Assume that path A contains 3 sub-links and path B contains 4 sub-links. The collected information includes the delay (L), bandwidth (B) and utilization (U) of each sub-link. For example:

[0111] Sublink 1 of path A: L = 10 ms, B = 100 Mbps, U = 60%;

[0112] Sublink 2 of path A: L = 15 ms, B = 80 Mbps, U = 50%;

[0113] Sublink 3 of path A: L = 8 ms, B = 90 Mbps, U = 70%;

[0114] Sublink 1 of path B: L = 12 ms, B = 95 Mbps, U = 65%;

[0115] Sublink 2 of path B: L = 20 ms, B = 85 Mbps, U = 55%;

[0116] Sublink 3 of path B: L = 18 ms, B = 75 Mbps, U = 60%;

[0117] Sublink 4 of path B: L=14 ms, B=110 Mbps, U=40%.

[0118] Step S2: Calculate the comprehensive weight;

[0119] Based on the path status information collected above, the network administrator uses the following formula to calculate the comprehensive weight of the path:

[0120]

[0121] Among them, W p represents the comprehensive weight of path p; and They represent the weight coefficient, sublink delay, sublink bandwidth, sublink utilization and sublink impedance factor of the sublink numbered i under path p, respectively. p represents the total number of sub-links under path p; is the combined function of the sub-link states of the sub-link numbered i under the path p;

[0122] Among them, the combined function of the sub-link status and sub-link impedance factor Calculated by the following formulas:

[0123]

[0124] Among them, α p and β p is the adjustment parameter of path p, is the maximum sub-link delay of all sub-links under path p, is the adjustment parameter of the sub-link bandwidth utilization of the sub-link with sequence number i under path p;

[0125]

[0126] Among them, λ p It is a tuning parameter used to adjust the sensitivity of the algorithm, and its value is based on experience; is the average delay value of all sub-links under path p, which is obtained by measuring the delay of all sub-links and then calculating the average value of the sub-link delay; the adjustment parameter is the adjustment parameter of the impedance of the sub-link with sequence number i under path p.

[0127] Through the above formula, the network administrator can calculate the comprehensive weights of path A and path B respectively.

[0128] Step S3: Select the optimal path;

[0129] The network administrator selects a path with a lower weight for data transmission based on the calculated comprehensive weight. In this embodiment, assuming that the comprehensive weight of path A is 0.85 and the comprehensive weight of path B is 0.90, path A is selected as the current optimal path.

[0130] Step S4: dynamically adjust the path selection strategy;

[0131] While data transmission is in progress, the network administrator monitors the load of path A in real time. If the load factor of path A is F p If it gradually increases and exceeds a certain threshold, the system automatically switches to path B for data transmission to ensure network load balancing.

[0132] Step S5: dynamic monitoring and execution;

[0133] The network administrator sets up a real-time monitoring module to continuously execute steps S1 to S4 to dynamically monitor the path status in the SRv6 network and automatically adjust the routing selection strategy to achieve multi-path load balancing control of the network. Specific embodiment three

[0135] The present invention also provides a multipath routing and load balancing control system for an SRv6 network, which is used to execute the multipath routing and load balancing control method for an SRv6 network, and the system includes:

[0136] Path monitoring module: used to monitor the link status information and load conditions of each path;

[0137] Path selection module: Based on the information provided by the monitoring module, it calculates the comprehensive weight of the path and selects the optimal path;

[0138] Load balancing module: used to dynamically adjust the path selection strategy to balance the load of each path;

[0139] Optimization solution module: based on the output of the path selection module, solve the optimal path selection strategy and generate a path routing table;

[0140] Strategy adjustment module: adjusts the path selection strategy in advance according to the load prediction results to avoid path congestion;

[0141] Feedback optimization module: monitors the effect of path selection in real time and continuously optimizes the path selection algorithm based on feedback information.

[0142] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A multipath routing selection and load balancing control method for SRv6 network, characterized in that: The method comprises the following steps: S1. Collect path status information in the SRv6 network, where the path status information includes at least sub-link delay, sub-link bandwidth, and sub-link utilization; S2. Calculate the comprehensive weight of each path based on the path status information. The comprehensive weight is calculated by the following formula: ， in, Indicates the path The comprehensive weight of , , , and Respectively represent the path The following serial number is The weight coefficient of the sub-link, sub-link delay, sub-link bandwidth, sub-link utilization and sub-link impedance factor, Indicates the path The total number of sub-links under For path The following serial number is A combined function of the sub-link states of the sub-links of ; S3, select the optimal path for data transmission according to the comprehensive weight, and record the load of the current path; S4. Based on the recorded load conditions, dynamically adjust the path selection strategy to balance the network load; S5, repeating steps S1 to S4, dynamically monitoring and executing multi-path routing selection and load balancing control, Wherein, in step S2, the combination function of the sub-link status is calculated by the following formula: ， in, For path The following serial number is The combined function of the sub-link states of the sub-links of and For path The adjustment parameters of For path The maximum sub-link delay of all sub-links under For path The following serial number is An adjustment parameter for the sub-link bandwidth utilization of the sub-link; Sublink Delay The values ​​are measured in real time. Adjustment parameters Used to control the function output range and adjust parameters Used to control sensitivity and adjust parameters Used to adjust the bandwidth utilization of sub-links and adjust parameters , Adjustment parameters and adjustment parameters The value of is affected by real-time delay and real-time bandwidth, and is predetermined through fitting experiments.

2. The multipath routing selection and load balancing control method for SRv6 network according to claim 1, characterized in that: In step S2, the sub-link impedance factor is calculated according to the following formula: ， in, represents the sub-link impedance factor, For path The adjustment parameter is used to adjust the sensitivity of the algorithm, and the value is based on experience. For path The average delay of all sub-links under is obtained by measuring the delay of all sub-links and then calculating the average of the sub-link delays; For path The following serial number is The impedance adjustment parameter of the sub-link is used to adjust the sensitivity of the algorithm to the path impedance, and the value is predetermined by fitting experiments.

3. The multipath routing selection and load balancing control method for SRv6 network according to claim 2, characterized in that: Adjustment parameters , Adjustment parameters , Adjustment parameters , Adjustment parameters and adjustment parameters The initial value of is 1.

4. The multipath routing selection and load balancing control method for SRv6 network according to claim 1, characterized in that: In step S3, the optimal path is selected for data transmission according to the comprehensive weight value, including: The optimization model is solved, and the expression of the optimization model is: ， in, represents the multipath optimal load combination matrix; represents the multipath load delay intensity function; Represents the value range of all possible multipath load combination matrices; represents the minimization solution operation, It means that within the range of all possible multipath load combination matrices, the multipath load combination matrix that makes the path load delay intensity function obtain the minimum value is obtained as the multipath optimal load combination matrix ; Indicates the path p The transmission delay, represents the maximum allowed delay, Indicates the path p The load factor.

5. The multipath routing selection and load balancing control method for SRv6 network according to claim 4, characterized in that: The load factor is calculated by the following formula: ， in, Indicates the path p The load factor, For path p The total number of flows on ; and Respectively represent j The volume and duration of each flow.

6. The multipath routing selection and load balancing control method for SRv6 network according to claim 4, characterized in that: In step S5, dynamic multi-path routing selection and load balancing control include: S51, performing real-time load monitoring on each path and calculating the load change rate of the path; S52, adjusting the flow transfer strategy based on the load change rate so that the load tends to be balanced; S53: If it is detected that the load change rate of a certain path exceeds a threshold, a path with a lower load is preferentially selected for data transmission.

7. The multipath routing selection and load balancing control method for SRv6 network according to claim 6, characterized in that: In step S51, the load change rate is calculated by the following formula: ， in, Indicates the load change rate, Indicates the path p The initial load factor, Indicates the time interval The path after p The current load factor.

8. The multipath routing selection and load balancing control method for SRv6 network according to claim 6, characterized in that: Step S52 includes: S521, calculate the load factors of all available paths, select the path with the largest load factor as the outbound target, select the path with the smallest load factor as the inbound target, and transfer the traffic; S522, looping through step S521, gradually transferring the traffic on the path with a high load change rate to the target path until the load balancing condition is met.

Images