PUEGM: A Method of User Revenue Selection Based on a Publisher-User Evolutionary Game Model for Mobile Crowdsensing
Abstract
:1. Introduction
- (1).
- Publisher-User Evolutionary Game Model (PUEGM): We modeled benefits of both publishers and users as evolutionary games, and deduced the evolutionary stability strategy (ESS), which can determine the choice of user revenue strategy.
- (2).
- Analyzing the impact of user data quality on its revenue in MCS, the ultimate user revenue is closely related to the data quality, and it is difficult to guarantee user reasonable revenue in the absence of data quality assessment. We analyzed the impact of user data quality on its revenue. The results can be applied to the user revenue selection under normal circumstances.
- (3).
- User data quality assessment algorithm. We propose an algorithm to remove the quality of low and abnormal data. At the same time, weighting factors are introduced to meet the different data quality concerns. We evaluated and compared the rationality of the proposed algorithm for classifying data quality and recognizing abnormal data.
- (4).
- Finally, as far as we know, this is the first time that the user revenue strategy is selected from the point of view of removing low and abnormal data quality in MCS data quality assessment. We think this is a tentative work for MCS research.
2. Related Work
2.1. Game Theory on MCS
2.2. Quality-Oriented MCS
3. Publisher-User Evolutionary Game Model (PUEGM) and Problem Statement
3.1. Model Assumptions
- (1).
- The game players are divided into two groups, namely, publishers and users. Publishers can select different incentive mechanisms for the same task to be released according to the actual needs, and users can obtain different benefits according to the incentive mechanism of publishers (multi-selection of player strategies.).
- (2).
- User’s data quality selection strategy and publisher’s incentive strategy change dynamically over time (strategy dynamics)
- (3).
- Publishers’ revenue is directly proportional to the size of incentive mechanism and users’ revenue is directly proportional to the quality of data (revenue relevance).
3.2. Model Construction
- (1).
- N = (Nu, Np) is the participants of both evolutionary games, where Nu denotes the user, and Np denotes the publisher.
- (2).
- W = (uw, pw) is the strategy space of both sides, where uw = {uw1, uw2, …, uwn} denotes users’ data quality optional strategy sets, and pw = {pw1, pw2, …, pwn} denotes publishers’ task-motivated optional strategy sets.
- (3).
- R = (d, q) is the game belief set, where di denotes the probability that publishers choose pwi, and qj denotes the probability that users select uwj.
- (4).
- B = (Bu, Bp) is the set of the revenue functions of both sides, where Bu denotes the revenue of the user and Bp denotes the revenue of the publisher.
3.3. Evaluate the Quality of User Data
3.4. Analyze the User Optimal Revenue Strategy
3.5. Problem Formulation
4. Algorithm Introduction and User Revenue Analysis
4.1. Data Quality Assessment Algorithm
Algorithm 1: Implementation of user data quality assessment |
Input: User data |
Output: Sorted the user data quality |
1: Initialization |
2: Determine the user data quality assessment indexes cj |
3: for i = 1,2,… ,n, j = 1,2,…,m do |
4: Calculate gi,j |
5: Judge if cj belongs to the cost index or benefit index |
6: if cj ∈ Ccost |
7: Calculate the error value of data by Equation (1) |
8: else |
9: Calculate the error value of data by Equation (2) |
10: end if |
11: Calculate by Equation (3) |
12: if |
13: Calculate and by equations (4) and (5) |
14: Sort the user data using Equation (6) |
15: else |
16: Remove errors and low-quality data |
17: end for |
4.2. User Revenue Optimal Strategy Selection Algorithm
Algorithm 2: Implementation of user optimal revenue strategy |
Input: Publisher-user game tree |
Output: Users’ optimal revenue strategy |
1: Initialization |
2: Build users’ type space collection and optional strategy space collection |
3: Select reasonable users’ strategy with probability , where |
4: Calculate |
5: Calculate and by Eqs. 8 and 10 |
6: Establish users’ replication dynamic equation u(q) and calculate the evolution equilibrium point |
7: Construct a Jacobian matrix to analyze the stability of the equilibrium point and obtain a stable equilibrium solution |
8: Output users’ revenue strategy |
9: End |
4.2.1. Analysis of Algorithm Time Complexity
4.2.2. Analysis of Spatial Complexity
4.3. Example Description
5. Performance Evaluation
5.1. Basic Simulation Setup
5.2. Experiment Results of Data Quality Assessment
5.3. Experiment Results of User Revenue Optimal Strategy
- (1)
- When d = 0 and q = 0, it means that publishers all provide strategy pwh, and users all provide strategy uwh. According to the simulation, we find that strategies selection has no change with the evolution time, as shown in Figure 4. The evolution result can be one of the stable states of the system, which also verifies that the user revenue can be maximized when publishers adopt a high incentive strategy and users adopt a high data quality strategy.
- (2)
- When d = 0.6 and q = 0.4, it means that publishers choose strategies pwl and pwh with a probability of (0.6,0.4) and users choose strategies uwl and uwh with a probability of (0.4,0.6). According to the simulation, we find that strategies selection has no change with the evolution time, as shown in Figure 5. The evolution result can be one of the stable states of the system, which verifies that this state is the equilibrium point.
- (3)
- When d = 1 and q = 1, it means that publishers all provide strategy pwl, and users all provide strategy uwl, as shown in Figure 6. According to the simulation, we find that strategies selection has no change with the evolution time, because publishers only provide a low incentive strategy, and users can’t get higher revenues even if they provide high quality data. Therefore, users only choose strategy uwl to ensure the optimal revenue of users.
- (4)
- When d = 0.4 and q = 0.3, it means that publishers choose strategies pwl and pwh with a probability of (0.4,0.6) and users choose strategies uwl and uwh with a probability of (0.3,0.7), as shown in Figure 7. After multiple games between the two players, we find that both sides of the game tend to F1 and users can choose strategy uwh to obtain the optimal revenue, because publishers have a higher probability to adopt a high incentive strategy and user is more likely to provide high quality data to maximize their revenues.
- (5)
- When d = 0.7 and q = 0.6, it means that publishers choose strategies pwl and pwh with a probability of (0.7,0.3) and users choose strategies uwl and uwh with a probability of (0.6,0.4), as shown in Figure 8. After multiple games between the two players, we find that the stable equilibrium solution of both sides of this game tends to F5 and users can choose strategy uwl to obtain the optimal revenue, because publishers have a higher probability to adopt a low incentive strategy and users are more likely to provide low quality data to avoid the loss of their revenues.
- (6)
- When d = 0.4 and q = 0.6, it means that publishers choose strategies pwl and pwh with a probability of (0.4,0.6) and users choose strategies uwl and uwh with a probability of (0.6,0.4), as shown in Figure 9. Although, the stable equilibrium solution of both sides of the game tends to F4 at the beginning, it tends to F5 eventually. The reason for this result is that the values of d and q are very close to F4 at the beginning, which leads to a balance between two players in a short time. However, in F4, the values of d and q can only be the initial equilibrium state, and users are not willing to provide high quality data. Therefore, users will eventually choose strategy uwl to achieve the best revenue.
- (7)
- We find that the stable equilibrium solutions of the two players are different in various initial states, as shown in Figure 10. When the value is 1, it means that both sides of the game will tend to F5 and users can get the best revenue by choosing strategy uwl. When the value is 0, it means that both sides of the game will tend to F1 and users can get the best revenue by choosing strategy uwh. When the value is 0.5, it means that the game will tend to F4 and users select strategies uwl and uwh with a probability of (0.4,0.6) to achieve the best revenue. We find that the higher quality of user data, the more stable equilibrium solution tends to F1, when publishers strategy is unchanged.
6. Conclusions
Author Contributions
Funding
Conflicts of Interest
References
- Zhang, X.L.; Yang, Z.; Sun, W.; Liu, Y.H.; Tang, S.H.; Xing, K.; Mao, X.F. Incentives for Mobile Crowd Sensing: A Survey. IEEE Commun. Surv. Tut. 2016, 18, 54–67. [Google Scholar] [CrossRef]
- Alvear, O.; Calafate, C.T.; Cano, J.C.; Manzoni, P. Crowdsensing in Smart Cities: Overview, Platforms, and Environment Sensing Issues. Sensors 2018, 18, 460. [Google Scholar] [CrossRef] [PubMed]
- Wang, X.; Zhang, J.B.; Tian, X.H.; Gan, X.Y.; Guan, Y.F.; Wang, X.B. Crowdsensing-Based Consensus Incident Report for Road Traffic Acquisition. IEEE Trans. Intell. Transp. Syst. 2018, 19, 2536–2547. [Google Scholar] [CrossRef]
- Nie, J.; Luo, J.; Xiong, Z.; Niyato, D.; Wang, P. A Stackelberg Game Approach Toward Socially-Aware Incentive Mechanisms for Mobile Crowdsensing. IEEE Trans. Wirel. Commun. 2019, 18, 724–738. [Google Scholar] [CrossRef]
- Xiao, L.; Chen, T.H.; Xie, C.X.; Dai, H.Y.; Poor, H.V. Mobile Crowdsensing Games in Vehicular Networks. IEEE Trans. Veh. Technol. 2018, 67, 1535–1545. [Google Scholar] [CrossRef]
- Pouryazdan, M.; Fiandrino, C.; Kantarci, B.; Soyata, T.; Kliazovich, D.; Bouvry, P. Intelligent Gaming for Mobile Crowd-Sensing Participants to Acquire Trustworthy Big Data in the Internet of Things. IEEE Access 2017, 5, 22209–22223. [Google Scholar] [CrossRef]
- Xiao, L.; Li, Y.D.; Han, G.A.; Dai, H.Y.; Poor, H.V. A Secure Mobile Crowdsensing Game With Deep Reinforcement Learning. IEEE Trans. Inf. Forensics Secur. 2018, 13, 35–47. [Google Scholar] [CrossRef]
- Yang, S.; Wu, F.; Tang, S.J.; Gao, X.F.; Yang, B.; Chen, G.H. On Designing Data Quality-Aware Truth Estimation and Surplus Sharing Method for Mobile Crowdsensing. IEEE J. Sel. Areas Commun. 2017, 35, 832–847. [Google Scholar] [CrossRef]
- Song, B.Y.; Shah-Mansouri, H.; Wong, V.W.S. Quality of Sensing Aware Budget Feasible Mechanism for Mobile Crowdsensing. IEEE Trans. Wirel. Commun. 2017, 16, 3619–3631. [Google Scholar] [CrossRef]
- Yang, G.; He, S.B.; Shi, Z.G.; Chen, J.M. Promoting Cooperation by the Social Incentive Mechanism in Mobile Crowdsensing. IEEE Commun. Mag. 2017, 55, 86–92. [Google Scholar] [CrossRef]
- Peng, D.; Wu, F.; Chen, G.H. Data Quality Guided Incentive Mechanism Design for Crowdsensing. IEEE Technol. Mobile Comput. 2018, 17, 307–319. [Google Scholar] [CrossRef]
- Liu, S.Z.; Zheng, Z.Z.; Wu, F.; Tang, S.J.; Chen, G.H. Context-Aware Data Quality Estimation in Mobile Crowdsensing. In Proceedings of the INFOCOM 2017-IEEE Conference on Computer Communications, Atlanta, GA, USA, 1–4 May 2017; pp. 1–9. [Google Scholar]
- Wang, E.; Yang, Y.J.; Lou, K.H. User selection utilizing data properties in mobile crowdsensing. Inf. Sci. 2019, 490, 210–226. [Google Scholar] [CrossRef]
- Restuccia, F.; Ghosh, N.; Bhattacharjee, S.; Das, S.K.; Melodia, T. Quality of Information in Mobile Crowdsensing: Survey and Research Challenges. ACM Trans. Sensor Netw. 2017, 13, 34. [Google Scholar] [CrossRef]
- Simon, H.A.J.D. Organization, Theories of Bounded Rationality. 1972, 1, 161–176. Available online: https://www.researchgate.net/publication/242617110_Theories_of_Bounded_Rationality (accessed on 6 May 2019).
- Liu, X.; Ota, K.; Liu, A.F.; Chen, Z.G. An incentive game based evolutionary model for crowd sensing networks. Peer Peer Netw. Appl. 2016, 9, 692–711. [Google Scholar] [CrossRef]
- Chen, Y.; Liu, Y.A.; Shao, Z.H. Data quality evaluation method based on the error-eliminating decision-making theory. J. Harbin Eng. Univ. 2018, 39, 172–177. (In Chinese) [Google Scholar]
- Yang, Y.; Che, B.C.; Zeng, Y.; Cheng, Y.; Li, C.Y. MAIAD: A Multistage Asymmetric Information Attack and Defense Model Based on Evolutionary Game Theory. Symmetry 2019, 11, 215. [Google Scholar] [CrossRef]
- Meng, S.; Wang, Y.; Jiao, L.; Miao, Z.Y.; Sun, K. Hierarchical evolutionary game based dynamic cloudlet selection and bandwidth allocation for mobile cloud computing environment. IET Commun. 2019, 13, 16–25. [Google Scholar] [CrossRef]
- Khan, A.A.; Abolhasan, M.; Ni, W. An Evolutionary Game Theoretic Approach for Stable and Optimized Clustering in VANETs. IEEE Trans. Veh. Technol. 2018, 67, 4501–4513. [Google Scholar] [CrossRef]
- Ruan, N.; Gao, L.; Zhu, H.J.; Jia, W.J.; Li, X.; Hu, Q. Toward Optimal DoS-resistant authentication in Crowdsensing Networks via Evolutionary Game. In Proceedings of the 36th IEEE International Conference on Distributed Computing Systems, ICDCS 2016, Nara, Japan, 27–30 June 2016; pp. 364–373. [Google Scholar]
- Huang, H.R.; Cai, K. A method of fuzzy multiple attribute decision making based on the error-eliminating theory. J Intell. Fuzzy Syst. 2016, 31, 2119–2127. [Google Scholar] [CrossRef]
- Marjanovic, M.; Skorin-Kapov, L.; Pripuzic, K.; Antonic, A.; Zarko, I.P. Energy-aware and quality-driven sensor management for green mobile crowd sensing. J. Netw. Comput. Appl. 2016, 59, 95–108. [Google Scholar] [CrossRef]
- Lin, Y.M.; Wang, H.Z.; Zhang, S.; Li, J.Z.; Gao, H. Efficient quality-driven source selection from massive data sources. J. Syst. Softw. 2016, 118, 221–233. [Google Scholar] [CrossRef]
- Cassese, D. Replicator Equation and the Evolution of Cooperation on Regular Communities; Springer International Publishing: Cham, Switzerland, 2018; pp. 869–880. [Google Scholar]
- Smith, J.M.; Price, G.R. The Logic of Animal Conflict. Nature 1973, 246, 15–18. [Google Scholar] [CrossRef]
- Ahmad, S.; Tahar, R.M.; Muhammad, S.F.; Munir, A.B.; Rahim, R.A. Application of system dynamics approach in electricity sector modelling: A review. Renew. Sustain. Energy Rev. 2016, 56, 29–37. [Google Scholar] [CrossRef]
- Resendis-Antonio, O. Jacobian Matrix. Encycl. Syst. Biol. 2013, 46, 1061–1062. [Google Scholar]
- Firdose, S.; Lopes, L.; Moreira, W.; Sofia, R.; Mendes, P. CRAWDAD Dataset Copelabs/Usense (v. 2017-01-27). Available online: https://crawdad.org/copelabs/usense/20170127 (accessed on 6 May 2019).
- Zhong, H.M.; Miao, C.Y.; Shen, Z.Q.; Feng, Y.H. Comparing the learning effectiveness of BP, ELM, I-ELM, and SVM for corporate credit ratings. Neurocomputing 2014, 128, 285–295. [Google Scholar] [CrossRef]
Equilibrium Point | Determinant and Trace | Det | Trace |
---|---|---|---|
+ | - | ||
+ | + | ||
+ | + | ||
+ | - |
Index | Quantification Range | |
---|---|---|
1. Response time ()/ min | [10,90] | |
2. Distance ()/ m | [0,5000] | |
3. Data integrity () | STATIONARY | [0.7,0.9] |
WALKING | [0.5,0.7) | |
RUNNING | [0.3,0.5) | |
4. Data reliability () | QUIET | [0.75,0.9] |
NORMAL | [0.6,0.75) | |
ALERT | [0.45,0.6) | |
NOISY | [0.3,0.45) |
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Share and Cite
Shao, Z.; Wang, H.; Feng, G. PUEGM: A Method of User Revenue Selection Based on a Publisher-User Evolutionary Game Model for Mobile Crowdsensing. Sensors 2019, 19, 2927. https://doi.org/10.3390/s19132927
Shao Z, Wang H, Feng G. PUEGM: A Method of User Revenue Selection Based on a Publisher-User Evolutionary Game Model for Mobile Crowdsensing. Sensors. 2019; 19(13):2927. https://doi.org/10.3390/s19132927
Chicago/Turabian StyleShao, Zihao, Huiqiang Wang, and Guangsheng Feng. 2019. "PUEGM: A Method of User Revenue Selection Based on a Publisher-User Evolutionary Game Model for Mobile Crowdsensing" Sensors 19, no. 13: 2927. https://doi.org/10.3390/s19132927
APA StyleShao, Z., Wang, H., & Feng, G. (2019). PUEGM: A Method of User Revenue Selection Based on a Publisher-User Evolutionary Game Model for Mobile Crowdsensing. Sensors, 19(13), 2927. https://doi.org/10.3390/s19132927