Resource scheduling method for concurrent competition requests in quantum network

Abstract

The invention provides a resource scheduling method for concurrent competition requests in a quantum network. The method comprises the following steps: establishing a routing path of a single request; calculating a congestion factor of each request routing path; establishing a demand-oriented multi-request purification scheduling method; and establishing a network throughput evaluation method. According to the method, a new demand-oriented purification mode is provided, and excessive consumption of bottleneck link resources in the network can be avoided while the fidelity constraint is met. More network resources are released by designing congestion factors to determine the purification order of each request. Compared with a traditional resource allocation scheme, the demand-oriented resource scheduling DRS method provided by the invention can realize higher network throughput under a similar resource conversion rate.

Description

Technical Field

[0001] The present invention belongs to the field of network protocols and quantum technology, and in particular relates to a resource scheduling method for concurrent competing requests in a quantum network. Background Art

[0002] In recent years, as a novel network architecture, quantum networks with quantum physical properties are profoundly affecting the world. Quantum networks can support a variety of new quantum applications, including quantum key distribution, distributed quantum computing, quantum clock synchronization, etc. A key prerequisite for realizing these functions is that the two parties requesting the service establish an end-to-end entangled connection. Due to link loss, it is difficult for two distant quantum nodes, namely source-destination (SD) nodes, to directly share entangled quantum bit pairs. In order to establish end-to-end entanglement, entanglement exchange technology (which is well known to those skilled in the art) is required. However, the entanglement distribution process based on entanglement exchange causes a decrease in the quality of the entangled connection. Since quantum network applications have strict requirements on the quality of the entangled quantum bits used, end-to-end entanglement fidelity is widely used to evaluate the quality of entangled connections.

[0003] At present, in order to meet the fidelity constraints of quantum applications, entanglement purification technology is usually introduced in the entanglement distribution process. It establishes high entanglement fidelity pairs by consuming low entanglement fidelity pairs. In this way, the fidelity of the established end-to-end entangled connection is guaranteed, but a certain amount of network entanglement resources are sacrificed. In addition to the additional resource consumption caused by purification, the demand competition between multiple SD pairs in the network and the probabilistic nature of various quantum operations have exacerbated the scarcity of network entanglement resources. Therefore, in a quantum network with limited resources, how to effectively utilize entanglement resources for concurrent competing requests becomes a key issue, which can be called a resource scheduling problem.

[0004] Although existing research has proposed some solutions to the resource scheduling problem, it has not fully considered the competition problem between concurrent requests from the network level. Due to the limited link capacity, concurrent requests in the network will inevitably introduce bottleneck links. In addition, since the purification operation sacrifices the entanglement pairs, the capacity of some links will be further limited. The emergence of these bottleneck links makes some demands unsatisfactory, ultimately leading to a decrease in network throughput. Therefore, it is necessary to design a resource scheduling solution under concurrent competing requests based on the overall network needs. Summary of the invention

[0005] Based on the above considerations, the present invention provides a resource scheduling method for concurrent competing requests in a quantum network, which specifically includes the following steps:

[0006] Step 1: Establish a routing path for a single request;

[0007] To preserve resources as much as possible, given the fidelity threshold of an application request, a path search method with end-to-end fidelity as the metric is selected;

[0008] Step 2: Calculate the congestion factor of each request routing path;

[0009] Define the congestion factor η as follows:

[0010]

[0011] In the formula, (v i , v j ) is the quantum link formed by the i-th quantum node v i and the j-th quantum node v j . η(v i , v j ) represents the congestion factor of the quantum link (v i , v j ). R k is the k-th parallel request in the network, and the total number of requests is m. R k (v i , v j ) represents that the k-th parallel request contains the link (v i , v j ). The magnitude of the congestion factor represents the ideal network resource demand of each link. When purifying the determined routing path for each request, sort according to the congestion factor and preferentially purify those "uncongested" links;

[0012] Step 3: Establish a demand-oriented multi-request purification scheduling method;

[0013] Assume that the form of the quantum channel is a phase-damping channel. In a multi-hop quantum network, for a Bell state, as entanglement swapping proceeds, the entanglement fidelity F n of the obtained S-D pair satisfies the following decay law:

[0014]

[0015] In the formula, n represents the number of path hops, the channel coefficient q = 1 - p, p is the phase-damping coefficient, and q i is the q value of the i-th phase-damping channel. From Equation (2), it can be seen that when the environmental noise has a large impact on the entangled pair, the fidelity drops sharply and the end-to-end fidelity cannot be guaranteed. To meet the fidelity constraint of quantum applications, entanglement purification technology needs to be introduced. Since the purification operation is also a probabilistic operation, to increase the success probability of the purification operation, a purification model with multiple attempts is adopted. Assume that the initial fidelity of the quantum link is f0, and the fidelity after purification is Here P pois the success probability of a single purification, and K is the number of purification attempts; for Bell pairs f1, f2, P po (f1, f2) = f1f2 + (1 - f1)(1 - f2), P po (f1, f2) is the success probability of a single purification of Bell pair f1, f2;

[0016] A demand - oriented purification mode changes the purification order of links by introducing a congestion factor. The purification operations will be executed in sequence according to the magnitude of the congestion factor. When the end - to - end fidelity meets the constraints, the purification operation is completed; this purification mode will achieve the goal of releasing bottleneck link resources;

[0017] Step 4: Establish a network throughput evaluation method;

[0018] To further evaluate the impact of DRS on network throughput, the network throughput is defined as the expected number of end - to - end entangled connections that meet the fidelity constraints and are finally established for all requests in the network; first, clarify the number of entangled connections T(v i , v j ) that can be established for any link (v i , v j ) in the network, which is divided into two cases: with and without purification:

[0019]

[0020] In the formula, N e is the number of entangled pairs consumed by the purification operation, and it satisfies N e = 2K. The capacity c(v i , v j ) is the maximum number of link - level entanglements that can be generated for any quantum link (v i , v j ). Assuming the success rate of entanglement swapping is l, then for the k - th request, that is, the k - th S - D pair (s k , d k ) with the source node s k , and the destination node d k , the path throughput is T(s k , d k ) = min[T(v i , v j )]l n-1 ; The network throughput T net is calculated as Before each request response, the remaining resources in the network need to be updated, and then the evaluation of network throughput is completed;

[0021] Through the entanglement swapping technology, the network throughput obtained is the number of end-to-end entanglement connections finally established in the network that meet the fidelity constraint, and further serves the upper-layer application.

[0022] In step 3 of a specific embodiment of the present invention, the number of purification attempts K = 3.

[0023] In an embodiment of the present invention, the Python simulation software is used to evaluate the method proposed in claim 1 under the Waxman random topology model. To further evaluate the performance of the DRS algorithm, the resource conversion rate Rcr is defined: Rcr is the ratio of the expected number of end-to-end entanglement connections that finally meet the fidelity constraint to the number of consumed entanglement pairs in the network, that is, Rcr = T net / N ep ; where N ep is the total number of consumed entanglement pairs in the network; the higher the Rcr value, the better the algorithm utilizes the network resources.

[0024] In the present invention, a resource scheduling method for concurrent competing requests in a quantum network is proposed. First, due to the existence of concurrent requests, the resource demands of different links in the network are different. Therefore, a congestion factor is introduced to evaluate the resource demand of each link. Second, to establish entanglement connections that meet the fidelity constraint and obtain higher network throughput, for the routing path of each request, the purification operation will be carried out in sequence according to the size of the congestion factor, avoiding excessive consumption of bottleneck links to a certain extent. Finally, numerical simulation is used to evaluate the performance of the algorithm proposed in the present invention.

[0025] The beneficial effects of the present invention are:

[0026] 1. The present invention proposes a new demand-oriented purification mode, which can avoid excessive consumption of bottleneck link resources in the network while meeting the fidelity constraint.

[0027] 2. Considering the competition relationship of concurrent requests in the network, the present invention designs a congestion factor to determine the purification order of each request, thereby releasing more network resources.

[0028] 3. Compared with the traditional resource allocation scheme, the demand-oriented resource scheduling (DRS) method proposed in the present invention can achieve higher network throughput under a similar resource conversion rate. Description of the Drawings

[0029] Figure 1 is a comparison diagram of purification operation modes, where Figure 1 (a) shows the greedy purification mode, Figure 1 (b) shows the hop-by-hop purification mode,Figure 1 (c) shows a demand - oriented purification mode;

[0030] Figure 2 is an example of dumbbell - shaped topology resource scheduling, where Figure 2 (a) shows the original topology, Figure 2 (b) shows the topology after the greedy purification mode, Figure 2 (c) shows the topology after the hop - by - hop purification mode, Figure 2 (d) shows the topology after the demand - oriented purification mode;

[0031] Figure 3 is the comparison of network throughput of three schemes under different scenarios, where Figure 3 (a) shows the throughput comparison under different fidelity thresholds, Figure 3 (b) shows the throughput comparison under different entanglement swapping success rates; Figure 3 (c) shows the throughput comparison under different numbers of SD pairs;

[0032] Figure 4 is the comparison of network resource conversion rates of three schemes under different scenarios, where Figure 4 (a) shows the resource conversion rate comparison under different fidelity thresholds, Figure 4 (b) shows the resource conversion rate comparison under different entanglement swapping success rates; Figure 4 (c) shows the resource conversion rate comparison under different numbers of SD pairs. Detailed implementation manner

[0033] The following combines the attached Figures 1 - 4 and embodiments to elaborate on the present invention in detail.

[0034] The present invention proposes a resource scheduling method for concurrent competing requests in a quantum network, and the method specifically includes the following steps:

[0035] Step 1: Establish a routing path for a single request;

[0036] Based on the establishment of entanglement connections between adjacent nodes in the quantum network, for a single long - distance request to establish an entanglement relationship, it is first necessary to determine a routing path. Since entanglement resources are scarce, the routing path determined for each request should maintain the highest possible end - to - end fidelity, thereby reducing the consumption of entanglement resources by purification operations. Therefore, in order to retain resources as much as possible, given the fidelity threshold of the application request, the present invention selects a path search method with end - to - end fidelity as the metric. A path with high end - to - end fidelity may not require entanglement purification operations, thereby releasing more entanglement resources, enabling more end - to - end entanglement connections in the network, and thus improving network throughput.

[0037] Step 2: Calculate the congestion factor of the routing path of each request;

[0038] To successfully establish an entangled connection for each long-distance request with fidelity guarantee, entanglement purification operations need to be completed on the selected routing path. In a quantum network, due to the existence of concurrent competing requests, the resource requirements of different links in the network are different, which will eventually lead to the emergence of bottleneck links. The purification operation will further reduce the capacity of these bottleneck links. Therefore, the present invention designs a demand-oriented purification mode. Before purification, a key issue is to determine the purification order of the links. The present invention will consider this issue from the network level.

[0039] Although the purification order of the links is designed for individual requests, the bottleneck links shared by multiple requests need to allocate resources for all requests. Therefore, it is necessary to introduce a parameter to evaluate the resource requirements of each link, and then determine the congestion degree of each link. The present invention defines the congestion factor η as follows:

[0040]

[0041] In the formula, (v i , v j ) is the quantum link formed by the i-th quantum node v i and the j-th quantum node v j . η(v i , v j ) represents the congestion factor of the quantum link (v i , v j ). R k is the k-th parallel request in the network, and the total number of requests is m. R k (v i , v j ) represents that the k-th parallel request contains the link (v i , v j ). The magnitude of the congestion factor represents the ideal network resource requirements of each link. When purifying the determined routing path for each request, the present invention will sort according to the congestion factor and preferentially purify those "uncongested" links, thereby releasing the link capacity of the bottleneck links to a certain extent.

[0042] Step 3: Establish a demand-oriented multi-request purification scheduling method;

[0043] After the congestion factors of the routing paths of each request are determined, a demand-oriented multi-request purification scheduling method can be established. Considering the environmental noise, in the present invention, it is assumed that the form of the quantum channel is a phase-damping channel. In a multi-hop quantum network, for a Bell state, as the entanglement swapping progresses, the entanglement fidelity F n of the finally obtained S-D pair satisfies the following attenuation law:

[0044]

[0045] wherein, n represents the path hop count, the channel coefficient q = 1 - p, p is the phase damping coefficient, and q i is the q value of the i-th phase damping channel. It can be seen from Equation (2) that when the environmental noise has a great influence on the entangled pairs, the fidelity drops sharply, and the end-to-end fidelity cannot be guaranteed. To meet the fidelity constraint of quantum applications, an entanglement purification technique needs to be introduced. Since the purification operation is also a probabilistic operation, to increase the success probability of the purification operation, the present invention adopts a purification model with multiple attempts. Assume that the initial fidelity of the quantum link is f0, and the fidelity after purification will be where P po is the success probability of one purification, K is the number of purification attempts, and in a specific embodiment of the present invention, K = 3 is taken. For Bell pairs f1, f2, P po (f1, f2) = f1f2 + (1 - f1)(1 - f2), and P po (f1, f2) is the success probability of one purification of the Bell pair f1, f2. As described above, the improvement of the fidelity brings about the consumption of additional entangled pairs, which in turn leads to the decrease of the network throughput. Therefore, the present invention proposes a new purification mode aimed at releasing the link capacity of the bottleneck link.

[0046] As Figure 1 shown, Figure 1 (a) and Figure 1 (b) show two commonly used purification modes at present. Figure 1 (a) is a purification method based on Greedy. To ensure the end-to-end fidelity, when the end-to-end fidelity is less than the fidelity threshold F th , all links are purified (the thick black line represents that this link is purified). Figure 1 (b) is a hop-by-hop purification mode. After each link is purified in sequence, a determination operation will be performed, and the purification will stop when the end-to-end fidelity meets the constraint. Figure 1 (c) is a demand-oriented purification mode proposed by the present invention. Compared with the hop-by-hop purification mode, the present invention changes the purification order of the links by introducing a congestion factor, and the purification operation will be performed in sequence according to the size of the congestion factor. When the end-to-end fidelity meets the constraint, the purification operation is completed. This purification mode will achieve the purpose of releasing the resources of the bottleneck link.

[0047] Step 4: Establish a network throughput evaluation method;

[0048] To further evaluate the impact of DRS on network throughput, the present invention gives the definition of network throughput. The present invention defines network throughput as the expected number of end-to-end entanglement connections that are finally established for all requests in the network and satisfy the fidelity constraint. First, clarify the number of entanglement connections T(v i ,v j ) that can be established for any link (v i ,v j ) in the network, which is divided into two cases: purification and non-purification:

[0049]

[0050] In the formula, N e is the number of entangled pairs consumed by the purification operation, and satisfies N e = 2K. The capacity c(v i ,v j ) is the maximum number of link-level entanglements that can be generated for any quantum link (v i ,v j ). Assuming that the success rate of entanglement swapping is l, then for the kth request, that is, the kth S-D pair (s k , d k ) with the source node s k and the destination node d k , the path throughput is T(s k , d k ) = min[T(v i , v j )]l n-1 . Therefore, the final network throughput T net can be calculated as At the same time, due to the limited link entanglement resources, the resources will be consumed after each request is responded to. Therefore, before each request response, it is necessary to update the remaining resources in the network, and then complete the evaluation of network throughput.

[0051] To better illustrate the scheduling process of DRS, Figure 2 a dumbbell-shaped topology resource scheduling example is given. The initial topology of the network is as shown in Figure 2 (a). Here, it is assumed that the phase damping coefficients of each link are the same, all 0.10 for easy calculation. In addition, it is assumed that the network fidelity threshold is 0.80, the entanglement swapping success rate is 0.95, the link capacity is 30, and requests 1 and 2 each need to establish 4 entanglement connections. Analyze the network throughput under three purification modes respectively. Taking the request (s1, d1) as an example, according to Equation (2), p1 = p2 = p3 = p4 = 0.10, so, q1 = q2 = q3 = q4 = 0.90. Therefore, the final end-to-end fidelity F4 = (1 + 0.9 2 *0.9 2 *0.92 *0.9 2 ) / 2 = 0.715233605. Since F4 < 0.80, the network needs to perform purification operations. Since the phase damping coefficients of each link are the same, only the purification situation of one link needs to be analyzed. Taking the link (s1, r1) as an example, first, according to Equation (2), the initial fidelity f0 of the link is calculated from the phase damping coefficient of the link as f0 = (1 + 0.9 2 ) / 2 = 0.905. Based on the purification equation the purified link fidelity is 0.98407, and then the updated phase damping coefficient p' = 0.016 (shown by the arrow in the figure) is obtained. Therefore, q' = 1 - p' = 0.984. Through the above analysis and calculation, for the four links of the request (s1, d1), substituting the updated q' into Equation (2), it is found that only two of the four links need to be purified to meet the fidelity constraint. At this time, the updated end-to-end fidelity F4 = (1 + 0.984 2 *0.984 2 *0.9 2 *0.9 2 ) / 2 = 0.8076 is higher than the fidelity threshold requirement of 0.80. However Figure 2 (b)'s greedy purification mode purifies all links, Figure 2 (c)'s hop-by-hop purification mode only purifies two links, but they both purify the bottleneck link (r1, r2) in the network, resulting in them only being able to meet the requirements of the request (s1, d1) ( Figure 2 (b) and Figure 2 (c)'s thick black solid lines). And the DRS algorithm proposed by the present invention, that is Figure 2 (d)'s demand-oriented purification mode purifies the links in sequence according to the size of the congestion factor, avoiding the bottleneck link (r1, r2) while purifying two links. Therefore, the network resources can simultaneously meet the requirements of (s1, d1) and (s2, d2) ( Figure 2 (d)'s black solid line and black dotted line), achieving higher network throughput.

[0052] Finally, through the entanglement swapping technology, the finally obtained network throughput is the number of end-to-end entanglement connections established in the network that meet the fidelity constraint, and further serves the upper-layer applications. Specific embodiments

[0054] The present invention uses Python simulation software to evaluate the proposed DRS algorithm under the Waxman random topology model. Regarding the comparison scheme, since the greedy and hop-by-hop purification modes do not have the ability of resource scheduling, the present invention adopts the resource allocation strategy proposed in the literature (J. Li, M. Wang, K. Xue, R. Li, N. Yu, Q. Sun, and J. Lu, "Fidelity-guaranteed entanglement routing in quantum networks," IEEE Trans. Commun. 70, 6748-6763, 2022) for concurrent requests, where the weight coefficients α = 0 and β = 1. In addition, to ensure fairness, the re-routing process is not considered in all three schemes.

[0055] In the evaluation, the initial phase damping coefficient of each quantum link satisfies p ∼ N(0.07, 0.01). In addition, the number of nodes in the network is defaulted to 40 and the link capacity is 50 in the simulation. The simulation runs 100 times, and the present invention gives the final average value. Additionally, to further evaluate the performance of the DRS algorithm, the concept of resource conversion rate Rcr is defined. Rcr is the ratio of the expected number of end-to-end entanglement connections that meet the fidelity constraint finally established to the number of consumed entanglement pairs in the network, that is, Rcr = T net / N ep . Where, N ep is the total number of consumed entanglement pairs in the network. The higher the Rcr value, the better the algorithm utilizes the network resources. This paragraph is used to illustrate the setting of parameters in the simulation evaluation, which is used to evaluate the effect of the present invention. The formula used is the definition of a parameter metric proposed by the present invention to evaluate the performance of the present invention.

[0056] To better evaluate the performance of the three schemes, the present invention analyzes the network throughput and resource conversion rate under different link fidelity thresholds, entanglement swapping success rates, and the number of SD pairs respectively. The method of controlling variables is adopted in the simulation. The default link fidelity threshold is 0.80, the entanglement swapping success rate is 0.90, and the number of SD pairs is 5. The total network demand is 100, which is randomly generated by each SD pair.

[0057] First, vary the link fidelity threshold from 0.70 to 0.90 and observe the changes in network performance. As Figure 3 (a) shows, the higher the fidelity threshold, the lower the network throughput. This is because a higher fidelity threshold requires more purification operations to ensure end-to-end fidelity. The large consumption of entanglement pairs results in a decrease in network throughput. The DRS algorithm can mitigate the "impact" brought by purification and has better throughput than the other two schemes. The resource conversion rates of the three schemes are shown in Figure 4(a). Although the greedy algorithm has more advantages in achieving end-to-end fidelity, a large number of purification operations result in the lowest resource conversion rate. The per-hop resource allocation scheme has the highest resource conversion rate, and the DRS algorithm ranks second. Compared with the per-hop scheme, the reason for the lower resource conversion rate of DRS is that while satisfying concurrent requests, the establishment of entangled connections in DRS also consumes resources. Moreover, due to the multiple-attempt purification scheme adopted in the present invention, the increase in throughput cannot fully match the consumption of entangled pairs. It is noted that when the fidelity threshold is 0.70, the resource conversion rate of DRS is basically the same as that of the per-hop scheme because at this time, the increase in throughput does not require excessive purification operations. When the fidelity threshold is 0.90, since the ability of DRS to support concurrent requests decreases, the resource conversion rates of DRS and the per-hop scheme are also very close. That is to say, the DRS algorithm proposed in the present invention can achieve higher network throughput at a resource conversion rate similar to that of traditional schemes.

[0058] Secondly, under different entanglement swapping success rates, Figure 3 (b) and Figure 4 (b) respectively depict the impacts of the three schemes on throughput and resource conversion rate. Since a higher entanglement swapping success rate means a larger number of entangled connections can be established between each SD pair, the throughputs of the three schemes are all increasing gradually. It can be seen that as the entanglement swapping success rate increases, the performance of DRS is the best. Because the DRS algorithm supports more concurrent requests, the increase in throughput per SD pair promotes the growth of network throughput. Since the increase in throughput is accompanied by the consumption of entangled pairs, there is no clear positive or negative correlation between the resource conversion rate and the entanglement swapping success rate. As the entanglement swapping success rate increases, the resource conversion rate of the DRS algorithm decreases compared with the per-hop scheme. This phenomenon is still attributed to the multiple-attempt purification scheme adopted to ensure the success rate of purification operations.

[0059] Finally, the present invention explores the impact of the number of SD pairs on the three schemes. In Figure 3 (c), the increase in the number of requests promotes the growth of network throughput because the idle resources in the network are more fully utilized. Compared with traditional schemes, the DRS algorithm has more obvious advantages in high-concurrency scenarios. In addition, when the number of SD pairs reaches 7, the network throughputs of each scheme have basically reached saturation, especially for the greedy algorithm. At the same time, the resource conversion rate of DRS can still remain at a stable level, as shown in Figure 4 (c).

[0060] Based on the above simulation results, it is proved that the DRS algorithm proposed in the present invention has superiority in resource scheduling, especially in high-concurrency scenarios. The results show that the DRS algorithm can achieve higher network throughput at a resource conversion rate similar to that of traditional schemes.

Claims

1. A resource scheduling method for concurrent competing requests in a quantum network, characterized in that, The method specifically includes the following steps: Step 1, establish the routing path for a single request; To preserve resources as much as possible, select a path search method with end-to-end fidelity as the metric under the given application request fidelity threshold; Step 2, calculate the congestion factor of each request routing path; Define the congestion factor η as follows: Wherein, (v i , v j ) is a quantum link formed by the i-th quantum node v i and the j-th quantum node v j . η(v i , v j ) represents the congestion factor of the quantum link (v i , v j ). R k is the k-th parallel request in the network, and the total number of requests is m. R k (v i , v j ) represents that the k-th parallel request includes the link (v i , v j ); the magnitude of the congestion factor represents the ideal network resource demand of each link; when purifying the routing path determined for each request, sort according to the congestion factor, and preferentially purify those "uncongested" links; Step 3, establish a demand-oriented multi-request purification scheduling method; Suppose the form of the quantum channel is a phase-damping channel; in a multi-hop quantum network, for Bell states, as entanglement swapping progresses, the entanglement fidelity \(F\) of the finally obtained S-D pair n obeys the following attenuation law: where n represents the path hop count, the channel coefficient q = 1 - p, p is the phase damping coefficient, and q i is the q value of the i-th phase damping channel; As known from Equation (2), when the environmental noise has a great impact on the entangled pairs, the fidelity drops sharply and the end-to-end fidelity cannot be guaranteed. To meet the fidelity constraints of quantum applications, entanglement purification technology needs to be introduced. Since the purification operation is also a probabilistic operation, to increase the success probability of the purification operation, a purification model with multiple attempts is adopted. Assume that the initial fidelity of the quantum link is f0 and the fidelity after purification is where P po is the success probability of a single purification attempt, and K is the number of purification attempts. For Bell pairs f1, f2, P po (f1, f2) = f1f2 + (1 - f1)(1 - f2), and P po (f1, f2) is the success probability of a single purification attempt for Bell pair f1, f2. A demand-oriented purification mode changes the purification order of the link by introducing the congestion factor, and the purification operations will be executed in sequence according to the magnitude of the congestion factor. When the end-to-end fidelity meets the constraints, the purification operation is completed; this purification mode will achieve the purpose of releasing the resources of the bottleneck link; Step 4, establish a network throughput evaluation method; To further evaluate the impact of DRS on network throughput, the network throughput is defined as the expected number of end-to-end entanglement connections that are finally established for all requests in the network and satisfy the fidelity constraint; first, clarify the number T(v i , v j ) of entanglement connections that can be established for any link (v i , v j ) in the network, which is divided into two cases: purification and non-purification: where N e is the number of entangled pairs consumed in the purification operation, and it satisfies N e = 2K, and the capacity c(v i , v j ) is the maximum number of link-level entanglements that any quantum link (v i , v j ) can generate; assuming that the success rate of entanglement swapping is l, then for the k-th request, that is, the k-th S-D pair (s k , d k ) with the source node s k , d k ), the path throughput is T(s k , d k ) = min[T(v i , v j )]l n-1 ; the network throughput T net is calculated as Before each request response, the remaining resources in the network need to be updated, and then the network throughput evaluation is completed; Through the entanglement swapping technology, the obtained network throughput is the number of end-to-end entanglement connections that finally meet the fidelity constraints in the network, and further serves the upper-layer application.

2. The resource scheduling method for concurrent competing requests in the quantum network according to claim 1, characterized in that, In Step 3, the number of purification attempts K = 3.

3. The resource scheduling method for concurrent competing requests in the quantum network according to claim 1, characterized in that, The method proposed in claim 1 is evaluated using Python simulation software under the Waxman random topology model. To further evaluate the performance of the DRS algorithm, the resource conversion rate Rcr is defined: Rcr is the ratio of the expected number of end-to-end entanglement connections that finally meet the fidelity constraint to the number of consumed entanglement pairs in the network, that is, Rcr = T net / N ep ; where N ep is the total number of consumed entanglement pairs in the network; the higher the Rcr value, the better the algorithm utilizes network resources.

Images