Progress in Signals and Telecommunication Engineering Hsing-Cheng Liu, Yao-Liang Chung 15 2 2026 https://doi.org/10.6025/pste/2026/15/2/64-87 https://www.dline.info/pste/fulltext/v15n2/pstev15n2_2.pdf The rapid expansion of wireless communication networks has necessitated a massive deployment of antenna infrastructure, yet the longitudinal and spatial temporal dynamics of antenna structure registrations remain poorly understood. This study addresses this gap by analyzing 124,811 Antenna Structure Registration (ASR) records in the United States from 1996 to 2012. Moving beyond simple aggregate counts, we employ an integrated analytical framework combining Seasonal Trend decomposition using Loess (STL), nonparametric trend testing (Mann Kendall, Sen's slope), multiple structural break detection algorithms (PELT, Bai Perron, Joinpoint, Bayesian), Markov switching regime analysis, and state level spatial temporal modeling. Results show a strong, statistically significant monotonic upward trend (+3.66 registrations/month) with weak seasonality. Crucially, the registration process is highly non-stationary, characterized by multiple structural breaks converging around 2011-2012 and two persistent latent regimes (low activity and high activity) with transition persistences exceeding 94%. We also identified 33 temporal anomalies, mostly clustered in the later observation period, indicating episodic acceleration rather than smooth growth. Conventional SARIMA forecasting demonstrated significant limitations (MAPE = 76.8%) in capturing this structural instability. Spatially, infrastructure deployment is highly concentrated, with Texas, California, and Florida accounting for over 20% of all registrations, though late period expansion was broadly distributed across leading states. Ultimately, this study demonstrates that antenna infrastructure evolution is a complex, regime dependent, and geographically heterogeneous process, establishing a foundational framework for future spatial temporal and regime aware infrastructure modeling.