Hierarchical Incentive Mechanism Design for Federated Machine Learning in Mobile Networks

Abstract

In recent years, the enhanced sensing and computation capabilities of Internet of Things (IoT) devices have opened the doors to several mobile crowdsensing applications. In mobile crowdsensing, a model owner announces a sensing task following which interested workers collect the required data. However, in some cases, a model owner may have insufficient data samples to build an effective machine learning model. To this end, we propose a Federated Learning based privacy preserving approach to facilitate collaborative machine learning among multiple model owners in mobile crowdsensing. Our system model allows collaborative machine learning without compromising data privacy given that only the model parameters instead of the raw data are exchanged within the federation. However, there are two main challenges of incentive mismatches between workers and model owners, as well as among model owners. For the former, we leverage on the self-revealing mechanism in contract theory under information asymmetry. For the latter, to ensure the stability of a federation through preventing free-riding attacks, we use the coalitional game theory approach that rewards model owners based on their marginal contributions. Considering the inherent hierarchical structure of the involved entities, we propose a hierarchical incentive mechanism framework. Using the backward induction, we first solve the contract formulation and then proceed to solve the coalitional game with the merge and split algorithm. The numerical results validate the performance efficiency of our proposed hierarchical incentive mechanism design, in terms of incentive compatibility of our contract design and fair payoffs of model owners in stable federation formation.

Incentive-Driven Federated Learning and Associated Security Challenges: A Systematic Review

“…[15]'s scheme is based on the non-cooperative game model, and their incentive model is designed based on the VCG mechanism, which is the opposite concept of the non-cooperative game approach. Their scheme outperforms the schemes proposed in [49], [62], [67].…”

“…In crowd-sensing, the presence of insufficient data samples makes it difficult to train efficient ML models. Lim et al [49] argue that the previously proposed schemes only consider a single MO which may not be always the case. They proposed a hierarchical incentive scheme for FL using contract and game theory to facilitate collaborative ML among multiple MOs.…”

Federated Learning for Internet of Things: A Comprehensive Survey

“…Here, users run classification and regression tree models at local devices and then offload the computed updates to the cloud for averaging protected by a security protocol built on the edge layer. Given a crowdsourcing/crowdsensing system, the work in [96] focuses on designing an incentive mechanism [97] by analyzing the interactions between the participating clients and the aggregator at an edge server to minimize learning costs. To be clear, each client selects a learning strategy for solving its local sub-problem to ensure desired accuracy with lowest participation costs, while the central server builds a utility function by averaging local updates to offer reward to the clients.…”

Radio Frequency Fingerprint Collaborative Intelligent Identification Using Incremental Learning

“…If the complete sensor observation data can be transmitted to the master node, the difference information between signals can be fully reflected, and the global optimization of RF-FI can be realized. Therefore, RF-FI based on distributed networks has strong application value [7]- [10].…”