EDGE AI FOR REAL-TIME TRANSACTION AUTHENTICATION IN IOT-BASED BANKING

Abstract

The ever-increasing demand for IoT devices in modern banking operations has invariably led to an age of real-time, personalized, and device-triggered financial interactions. However, all these transformations have brought a plethora of security threats, latency issues, and infrastructure limitations—especially when relying on centralized cloud architectures for transaction authentication. In this framework, we intend to introduce an innovative Edge AI solution proposed for real-time transaction authentication in IoT-based banking systems. This framework precipitates asset authentication from the central data node toward the edge of the network through the loading of lightweight, energy-efficient AI models that are directly incorporated within IoT devices such as smart ATMs, contactless POS systems, mobile wallets, and wearable payment tools.

Independent of this fact, the hybrid deep learning model with CNNs and LSTM units of this proposed architecture detects security anomalies and authenticates transactions with several biometric, behavioral, and contextual inputs. Optimization procedures, as much concerning the decision latency, became necessary, with quantization and pruning of the model architecture to constrain it within the computational and memory thresholds set by edge devices. Testing also heavily highlights benchmarking on a simulated testbed. The model uses signatures, transaction logs, and anomaly patterns as datasets. All tests have exhibited an excellent accuracy rate in the proximity of 94.7 while decreasing decision latency by about 45% as against traditional cloud models.

The argument continues that the aforementioned approach also enhances data privacy, as the system obviates the need to transport sensitive information concerning the individual over the internet. These results then reach the conclusion that Edge AI holds consistent value with respect to cloud-based security mechanisms while setting the foundation for the development of secure, scalable, intelligent financial IoT infrastructures. This paper ends with a thorough examination of issues like implementation scenarios, real-world viability, and major areas for future research direction in Edge AI for secure financial computing.