Mobile crowd sensing matching method for multi-platform cooperation

Abstract

The invention discloses a mobile crowd sensing matching method for multi-platform cooperation, and the method is realized through two stages: in the first stage, a Gal-Shapley algorithm is adopted to establish basic stable matching and obtain an initial matching result, and in the second stage, a TTC mechanism is introduced to carry out matching optimization on the initial matching result in the first stage, and cross-platform resource exchange is realized. According to the matching method, the limitation of NRMP static one-time matching is broken through, and continuous optimization is realized through a two-stage mechanism. On the basis of keeping initial matching stability, a TTC mechanism at the second stage supports worker exchange, so that the system efficiency is improved compared with that of a traditional scheme, the problem of benefit deviation caused by NRMP unidirectional preference is solved, constraints are improved through Pareto, and benefit balance of a platform and workers is ensured.

Description

TECHNICAL FIELD

[0001] The application relates to a mobile crowd sensing matching method, in particular to a mobile crowd sensing matching method for multi-platform cooperation. BACKGROUND

[0002] The resident matching program (NRMP) has achieved remarkable results in the field of medical resource allocation through the Gale-Shapley algorithm, but it still has several key defects when applied to a multi-platform mobile crowd sensing system. First, the static matching mechanism adopted by the system is only suitable for one-time matching scenarios and cannot adapt to the dynamic changes in resource demand in the MCS environment, such as real-time location changes of workers, adjustment of task priorities, or sudden task demand of platforms, and other dynamic factors. Second, the one-way preference strategy (hospital optimal or physician optimal) of NRMP will lead to systematic bias in the matching results, which contradicts the need to consider the interests of platforms, workers, and tasks in the multi-platform cooperation scenario of MCS. Third, the system lacks an effective resource reconfiguration mechanism and cannot be optimized and adjusted once the initial matching is completed, which will cause significant efficiency loss in the long-term operation of the MCS system. In addition, the centralized decision-making mode of NRMP is difficult to adapt to the distributed and decentralized characteristics of the MCS system, and the self-determination right of each platform cannot be guaranteed. Most importantly, the system does not consider the synergy effect across platforms and cannot solve the problems of resource fragmentation and negative externality caused by multi-platform competition. Therefore, it is urgent to establish a new multi-platform cooperation mechanism that can ensure the stability and rationality of the matching, solve the core problems of resource fragmentation, information asymmetry, and efficiency loss through collaborative resource allocation between platforms, and ultimately achieve the overall benefit improvement and sustainable development of the multi-platform MCS system. SUMMARY

[0003] The technical problem to be solved by the application is how to achieve efficient collaborative allocation of cross-platform resources in a multi-platform mobile crowd sensing system to solve the problems of allocation imbalance and low efficiency caused by competition for worker resources between platforms, and to improve the overall system efficiency while ensuring matching stability.

[0004] To solve the above technical problems, the application provides a mobile crowd sensing matching method for multi-platform cooperation, comprising the following steps:

[0005] S1. Establish a basic stable matching and obtain a matching result by using the Gale-Shapley algorithm.

[0006] S2. Optimize the matching result by pointing each worker to the platform he prefers most, and pointing each platform to the worker he initially matched.

[0007] S3. Find the closed loop in the worker and platform association, complete the final matching between the workers in the closed loop and the pointed platform, then these workers exit the matching process, if the corresponding platform's recruitment quota is zero, the platform also exits the matching, and the unselected platform keeps the original quota unchanged.

[0008] S4. Re-establish the pointing relationship in the remaining workers and platforms, repeat steps S2 and S3 until all workers are successfully assigned to platforms, and obtain the final matching result.

[0009] Further optimization of the above scheme, the specific steps of step S1 are as follows:

[0010] S1.1 Initialize the set of all workers participating in matching and the set of platforms, respectively establish a complete preference list for each set, and set all matching states to unmatching.

[0011] S1.2 Perform iterative matching, at the beginning of each round, identify all unmatching workers, these workers send matching requests to their most preferred platform in order of their preference; when the platform receives multiple requests, it selects the best workers according to its preference list and the upper limit of the number of workers to be recruited, and rejects other applicants, the rejected workers update their preference list and turn to the next best choice in the next round, until all workers are successfully matched or the remaining unmatching workers cannot find acceptable platforms in their preference list, and obtain a preliminary stable matching result.

[0012] Further optimization of the above scheme, the preference list construction includes at least worker contribution, platform offer, worker cost, and platform benefit parameters.

[0013] Further optimization of the above scheme, each platform sets a maximum number of matching workers.

[0014] Further optimization of the above scheme, the specific operation of step S1.2 is as follows:

[0015] Calculate each worker's preference list, calculate each platform's preference list, execute when there are idle workers and still platforms to apply, all idle workers send applications to their most preferred platform according to their preference list; if the platform's capacity is not up to the limit, compare the current applicant workers with the workers in the platform's waiting list, and reject the workers with the lowest preference; the rejected workers reapply to their next best choice; each platform updates its waiting list and rejects the remaining workers; return the preliminary matching result.

[0016] The application realizes continuous optimization through a two-stage matching framework. The first stage uses an improved Gale-Shapley algorithm to establish a basic stable matching, ensuring the basic satisfaction of each participant. The second stage introduces a TTC mechanism to realize cross-platform resource exchange and continuously optimize matching efficiency through dynamic negotiation. This hybrid strategy not only retains the stability advantage of classic matching algorithms but also enhances the adaptability of the system by introducing a dynamic adjustment mechanism. At the same time, the framework supports distributed decision-making, allowing each platform to maintain considerable autonomy while avoiding the loss of flexibility caused by excessive centralization. Ultimately, it significantly improves the overall operating efficiency and resource utilization of the multi-platform MCS system while ensuring matching stability. It solves the problem of interest bias caused by one-way preferences in NRMP. Through Pareto improvement constraints, it ensures the balance of platform and worker interests. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A multi-platform mobile crowd sensing matching example. DETAILED DESCRIPTION

[0018] The platform collaboration matching model is represented as where p is the set of platforms and w is the set of workers. Each platform has an upper limit q on the number of matched workers j , representing the maximum number of workers that can be matched by the platform. In the matching result, if a worker w i is successfully matched, then otherwise, if not matched, then For platform p j , the matching result is represented as where k≤q j , i.e. the actual number of matched workers does not exceed its capacity limit. In addition, the preference relationship can be expressed as: when , it means that the worker w i prefers platform p to its current matching result j ; similarly, means that platform p j prefers worker w to its current matching result i . This model ensures that each matched worker corresponds to only one platform and the number of matched workers of a platform is always within its capacity constraint

[0019] To better illustrate the matching model of the application, the application presents a matching example between the two sets p and w as Figure 1 where

[0020] In the matching model, both the platform and the worker are rational self-interested game players who make matching decisions according to their respective preferences. Each worker constructs a preference list The platforms are ranked in descending order, which can be expressed as:

[0021]

[0022] Similarly, the reference preference list of each platform is According to the demand satisfaction value in descending order, that is:

[0023]

[0024] The scheme of the present application adopts the classical Gale-Shapley algorithm in the first stage to realize the stable matching in the initial stage, and the execution process of the algorithm can be divided into the following steps: first, the system initializes the worker set w and the platform set p participating in matching, establishes a complete preference order list for each participant, and sets all matching states to be unmatched. In the iterative matching stage, the algorithm adopts a multi-round progressive matching mechanism: at the beginning of each round, the system identifies all the unmatched workers, and these workers will successively send matching requests to the most desirable platform according to their preference order; when the platform receives multiple requests, it will strictly screen according to its preference list and the upper limit of the number of recruited workers, and only keep the best workers and reject other applicants. The rejected workers will update their preference list and turn to the suboptimal choice in the next round. This process has a significant self-convergent property, and through repeated "application-screening-updating" cycles, the system state will gradually tend to be stable. Finally, the algorithm terminates when one of the following two conditions is met: all workers are successfully matched, or the remaining unmatched workers cannot find acceptable platforms in their preference lists. The theory proves that this mechanism must converge to a stable matching solution within a finite number of steps, which meets the individual rationality and non-blocking pair conditions, ensuring the fairness of the matching result and the stability of the system.

[0025] The TS-PCM first-stage matching algorithm based on the Gale-Shapley algorithm proposed by the present application is as follows.

[0026] Initialization: worker set W, platform set P, maximum number of recruited workers q of the platform j .

[0027] Input: worker contribution Platform bid Worker cost Platform benefit parameter λ j .

[0028] Output: platform-worker preliminary stable matching result M stable (W, P).

[0029] The specific steps are as follows:

[0030] 1. Calculate each worker preference list

[0031] 2. Calculate each platform preference list

[0032] 3. First stage: Gale-Shapley algorithm obtains a preliminary stable matching result

[0033] 4. When there are idle workers w i and there are still platforms p j that can apply:

[0034] 5. All idle workers send an application to their most preferred platform according to their preference list ;

[0035] 6. If the capacity q j of platform p j > 0:

[0036] 7. Compare the current application worker w i with the workers in the waiting list of platform p j , and reject the workers with the lowest preference;

[0037] 8. The rejected workers reapply for their next best choice;

[0038] 9. Each platform updates its waiting list and rejects the remaining workers;

[0039] 10. Return the preliminary matching result M stable (W, P).

[0040] In the first stage of the matching process, the matching mechanism based on the Gale-Shapley algorithm presents a significant dynamic evolution feature. Each participant, including workers and platforms, strictly follows its preset preference sequence to make selection decisions, but in the multi-round iterative matching process, participants often experience a cycle of rejection and reselection. This dynamic interaction mechanism, while ensuring the stability of the final matching result, introduces several issues worth noting: first, since the matching request is initiated in a single order, participants who initiate the request early may have priority in matching, and this timing advantage may lead to a compressed selection space for later participants; second, in the repeated rejection-reselection process, some participants may ultimately obtain a matching object that is far below their initial preference order, resulting in a significant "matching quality decay" phenomenon. More importantly, while this mechanism can guarantee the stability of the match, it may come at the expense of overall system efficiency, as some high-quality matching combinations that contribute more to platform efficiency may not be realized due to improper timing arrangements. This phenomenon reveals an essential feature of the algorithm: while a stable matching solution avoids the existence of blocking pairs, it is not necessarily Pareto optimal, and the final matching quality is largely subject to the initial request order and the arrangement of the preference list.

[0041] To address the suboptimal matching results that may arise from the traditional Gale-Shapley algorithm, an innovative optimization scheme is proposed.

[0042] In the second stage of the algorithm, after the initial matching is completed, all workers and platforms enter the matching optimization phase. First, each worker points to their most preferred available platform, including the currently matched platform, and each platform points to its initially matched worker. Due to the limited number of participants, the system must have at least one closed loop. When a closed loop is found, the workers within the closed loop will complete the final matching with the platforms they point to, and then these workers exit the matching process; at the same time, the number of available platforms decreases accordingly, and if the number is zero, the platform also exits the matching. Unselected platforms remain unchanged. Subsequently, a new pointing relationship is established among the remaining workers and platforms: workers choose the most preferred platform from those still available, and platforms continue to point to their initially matched workers, forming a new closed loop. This process is repeated until all workers are successfully assigned to platforms, and the algorithm terminates. The entire process ensures that stable matching is completed within a limited number of steps, and each subject participates in at most one closed loop.

[0043] The detailed process of the second-stage matching of TS-PCM based on the TTC algorithm optimization matching proposed by the application is as follows.

[0044] TS-PCM second-stage matching optimization algorithm

[0045] Input: platform-worker preliminary matching result Mstable (W, P).

[0046] Output: Platform-worker final matching result

[0047] The specific steps are as follows:

[0048] 1. Second stage: TTC optimal matching result

[0049] 2. Each worker w i ∈W, let it point to the platform with the highest preference

[0050] 3. All workers and platforms are marked as "not quit";

[0051] 4. When there are still unmatched workers, repeat the following steps:

[0052] 5. Each non-quitting worker w i points to the still available platform with its current most preferred;

[0053] 6. Each non-quitting platform p j points to the worker who has not quit among the workers matched in the first stage;

[0054] 7. Find a closed loop C in the worker-platform graph:

[0055] Such as w1→p1→w2→p2→…→w i →p k →w1;

[0056] 8. Process each pair of workers w i and platforms p j in the closed loop C:

[0057] 9. Let M(w i ) = p j (update the final matching);

[0058] 10. Worker w i is marked as "quit";

[0059] 11. q j = q j -1;

[0060] 12. If q j = 0, platform p j is marked as "quit";

[0061] 13. Return the final matching result

Claims

1. A mobile crowd-sensing and matching method for multi-platform collaboration, characterized in that, Includes the following steps: S1. Establish basic stable matchings and obtain matching results using the Gale-Shapley algorithm; S2. Optimize the matching results by directing each worker to their preferred available platform, and simultaneously directing each platform to its initially matched worker; S3. Find the closed loop in the worker and platform association, and make the workers in the closed loop finally match with the platform they are assigned to. Then these workers leave the matching process. If the recruitment quota of the corresponding platform is zero, the platform also leaves the matching. The unselected platforms keep their original quota unchanged. S4. Re-establish the pointing relationship among the remaining workers and platforms, and repeat steps S2 and S3 until all workers are successfully assigned to platforms, obtaining the final matching result.

2. The mobile crowd sensing and matching method for multi-platform collaboration as described in claim 1, characterized in that, The specific steps of step S1 are as follows: S1.1 Initialize all worker sets and platform sets participating in the matching, establish a complete preference sequence table for each set, and set all matching states to unmatched; S1.2 performs iterative matching. At the beginning of each round, all currently unmatched workers are identified. These workers send matching requests to their most preferred platforms in order of their preferences. When a platform receives multiple requests, it filters them based on its own preference list and the maximum number of workers it can recruit, retaining only the best-chosen workers and rejecting other applicants. Rejected workers update their preference lists and move to the next best option in the next round, until all workers are successfully matched or the remaining unmatched workers can no longer find an acceptable platform in their preference list, thus obtaining a preliminary stable matching result.

3. The mobile crowd sensing and matching method for multi-platform collaboration as described in claim 2, characterized in that: The preference sequence list construction includes at least the parameters of worker contribution, platform price, worker cost, and platform benefit.

4. The mobile crowd sensing and matching method for multi-platform collaboration as described in claim 2 or 3, characterized in that: Each platform has a maximum limit on the number of workers it can match.

5. The mobile crowd sensing and matching method for multi-platform collaboration as described in claim 4, characterized in that, The specific operation of step S1.2 is as follows: Calculate the preference sequence list for each worker; Calculate the preference sequence list for each platform; When there are idle workers and platforms are still available for application: All idle workers send applications to their most preferred platform according to their preference sequence list; if the platform's capacity is not at its limit, the currently applying worker is compared with the workers in the platform's waiting list, and the worker with the lowest preference is rejected; the rejected worker reapplies for its next best option; each platform updates its waiting list and rejects the remaining workers. Returns preliminary matching results.

Images