5G/ORAN Resiliency

O-RAN resiliency is the ability of an Open Radio Access Network to maintain reliable and predictable service even when parts of the system face failures, overloads, misconfigurations, or other disruptions. It is achieved through a combination of architectural design, redundancy strategies, automation, and intelligent control, rather than a single mechanism. As 5G networks evolve toward 6G, meeting stringent Service Level Agreements (SLAs) becomes increasingly challenging, making fault-tolerant networks more critical than ever. O-RAN’s softwarized architecture enables self-healing capabilities, and the integration of AI and machine learning within Near-Real-Time and Non-Real-Time RAN Intelligent Controllers allows networks to proactively detect issues, optimize performance, and recover from faults with minimal human intervention.

Open Radio Access Network (O-RAN), defined by the O-RAN Alliance, represents a major shift in mobile network design. By disaggregating the RAN into Radio Units (RU), Distributed Units (DU), and Centralized Units (CU), and integrating the RAN Intelligent Controller (RIC), O-RAN enables flexible deployment, cloud-native operation, and multi-vendor interoperability. This architecture is critical for supporting the high data rates, low latency, and scalable network requirements of next-generation 5G and future 6G applications.

At the same time, O-RAN’s disaggregated and virtualized architecture introduces significant resiliency challenges. Functions distributed across multiple locations make state management and synchronization complex, and open interfaces increase inter-component dependencies, meaning failures in one component can cascade across the network. Strict real-time constraints between the RU and DU leave little tolerance for latency or instability. Real-world deployments face failures such as CU and DU software crashes, signaling storms, resource contention, noisy neighbor effects from shared workloads, RIC malfunctions, fronthaul interface issues, and misconfigurations affecting network slices. These challenges make maintaining reliable service, meeting stringent SLAs, and ensuring continuous operation far more demanding than in traditional integrated RAN systems.

Some of the research problems are given below:

Problem	Description
Anomaly Detection & Prediction	Developing reliable methods to detect and predict anomalies in the network using telemetry data remains a significant challenge due to the complexity and scale of O-RAN deployments.
Explainable AI (xAI)	Ensuring transparency and operator trust in deep learning models, which often function as “black boxes,” is critical for safe and practical deployment in network control.
Resolution Strategy Planning	Designing effective and automated strategies for fault recovery is challenging, requiring careful planning to minimize downtime and maintain service continuity across distributed network components.

Research Scholars

Yaswanth Kumar L S, Somya Jain, Michael Suguna, Abdulla Ovais, Harshwardhan Gaikwad

Publications