📚Research

Academic Research

I build LLM-driven systems fine-tuning, inference pipelines, and agentic reasoning applied to the hard physical constraints of 6G wireless networks. Think of it as teaching language models to understand radio signals, city maps, and channel physics well enough to make decisions a telecom engineer would trust.

Google Scholar Profile

Publications

✓ Accepted • IEEE OJ-COMS 2026

Digital Twin-Guided AI Path Planning for Connectivity-Aware Mobility

Authors: Including Sai Teja Srivillibhutturu

IEEE Open Journal of the Communications Society (OJ-COMS) · 2026

Abstract: This journal article presents a comprehensive digital twin-guided AI framework for connectivity-aware mobility in 6G networks. Using Sionna-based ray-tracing simulations of mmWave propagation, the system trains AI agents to predict link quality along candidate routes and select paths that maximize sustained wireless connectivity. The extended work covers multi-environment generalization, beam management integration, and benchmarking against classical telecom planning baselines.

Digital Twin AI Path Planning 6G Mobility mmWave LLM Fine-Tuning Wireless Networks

IEEE Xplore (OJ-COMS) Read PDF Google Scholar

✓ Accepted • IEEE ICC 2026

CTMap: LLM-Enabled Connectivity-Aware Path Planning in Millimeter-Wave Digital Twin Networks

Authors: Including Sai Teja Srivillibhutturu

IEEE International Conference on Communications (ICC) - CQRM Workshop 2026

Abstract: CTMap is a novel framework that applies Large Language Model reasoning to connectivity-aware path planning in millimeter-wave digital twin networks. Given a city-scale OpenStreetMap graph and Sionna-simulated mmWave signal maps, CTMap prompts an LLM to evaluate route candidates by weighing predicted SINR, blockage probability, and handover cost producing paths that a standard GPS navigator would ignore. Fine-tuned on 10K+ labeled trajectory examples, CTMap outperforms both shortest-path and signal-greedy baselines on connectivity continuity metrics.

Large Language Models mmWave Networks Digital Twin Path Planning 6G Communications Transformer Fine-Tuning

Read Paper (arXiv) Google Scholar

Research Interests

🧠

LLM Fine-Tuning

Supervised fine-tuning and RLHF for domain-specific LLMs teaching transformers to reason about wireless environments and RF signal data.

🔗

Agentic AI Systems

Building LLM agents that take multi-step decisions querying simulation environments, evaluating options, and acting under physical constraints.

📡

6G & mmWave

Millimeter-wave propagation and channel modeling as the hard-domain challenge that drives the AI research.

🌐

Digital Twins for AI

Using Sionna ray-tracing as a synthetic data engine generating labeled wireless scenarios at scale to train LLM-based planners.

⚡

ML Pipelines

End-to-end data pipelines that convert telecom simulation outputs into structured features LLMs can consume bridging physics and language.

🗺️

Spatial Reasoning

Graph-based and map-aware reasoning for path planning grounding LLM decisions in real geometry via OpenStreetMap and city-scale network topology.

Research Experience

IEEE Publications

ICC

Top Conference

UTA

Research Lab

Graduate Research Assistant

The University of Texas at Arlington • Jun 2025 - Present

Core focus: LLM engineering applied to wireless networks. Built supervised fine-tuning pipelines using PyTorch and Hugging Face Transformers, generating 10K+ training examples from Sionna 6G digital twin simulations. Designed prompt schemas and tokenization strategies that encode mmWave channel maps, OSM road graphs, and signal quality scores into formats transformers can reason over enabling an LLM to plan routes the way a telecom engineer thinks about coverage. Resulted in two accepted IEEE publications at ICC 2026 (conference) and OJ-COMS (journal).