Privacy Partition: A Privacy-Preserving Framework for Deep Neural Networks in Edge Networks

Survey on Multi-Access Edge Computing Security and Privacy

“…In this approach, data or devices that include information are partitioned into various layers where different privacy preserving techniques can be applied effectively. Chi et al [161] introduced a novel technique called privacy partitioning which composed a trusted local partition and untrusted remote partition. The proposed method targets privacy preservation of deep learning classification tasks employed in mobile offloading processes.…”

“…Under these novel regulations, any web or hosting service intended to acquire data from any individual should draw their consent before initialization. Unauthorized CMDP based scheduling algorithm is proposed for task offloading process that ensures location and usage pattern privacy with optimum power consumption [160] A deep Post-Decision State (PDS) learning method for suggesting an optimal offloading strategy to the IoT device that preserve the location privacy of the UE from cyber-eavesdropping [161] Privacy partitioning method for deep learning classifications employing bipartite topology H-IoT and Big Data [162] Employed OPP and OJP methods for aggregating Laplacian random noise to the data for mis-informing the adversary. Privacy leakage probability is minimized Service Migration [158] Chaff service based approach for confusing the eavesdropper that preserves privacy of subscribers Authentication [163] AAKA protocol for preserving user privacy from identity protection IoT data aggregation [164] Three layer privacy preserving model for IoT enabled MEC deployments that employ a crypto system with homomorphic properties Mobility [165] MEC incorporated MSS model for preserving location privacy that demonstrate operational cost efficiency Smart Grid [166] Blockchain based edge computing model for Smart Grid Networks that guarantee user identity privacy GDPR [167] GDPR initiative handling and capturing of data is considered an offence.…”

“…Offloading-based privacy ensuring mechanisms are proposed in the literature [159], [160] that employ MDP-based scheduling for optimizing energy consumption with applied security. Privacy partitioning [161] is a novel approach utilized for applying different privacy mechanisms on various partitions. Privacy models and protocols proposed in [163], [164] could be adopted for the MEC system to ensure anonymity and identity of subscribers.…”

Artificial Intelligence for Securing IoT Services in Edge Computing: A Survey

“…(2) Privacy Partition. A practical method named privacy partition for privacy preservation in ML is presented in [65,66]. Privacy partition is a privacy-preservation framework for deep neural networks, and the basic structure of the framework is made up of a bipartite topology network and an interactive adversarial network [65].…”

“…e output of the last transformation in a trusted local computing context will be processed by a learning module. After that, the processed information will be the input of the first transformation layer in remote computing context [65]. Under the architecture of the edge network, privacy partition provides an optional choice to some centralized deep learning frameworks.…”