Dating back to as far as 1940, the US road and bridge infrastructure system has garnered quite the status for strategically connecting together half a continent. As monumental as the infrastructure's status, is its rate of deterioration, with the average bridge age coming at a disconcerting 50 years. Aside from visual inspection, a battery of non-destructive tests were developed to conduct structural fault assessment and detect laminations, in order to preemptively take preventive measures. The mainstream commercially favored test is the impulse time domain ground penetrating radar (GPR). An extremely short, high voltage pulse is used to visualize cross-sections of the bridge decks. While effective and it does not disturb traffic flow, impulse radar suffers from major drawbacks. The drawbacks are namely, its limited dynamic range and high cost of system manufacturing. A less prominent yet highly effective system, stepped frequency continuous wave (SFCW) GPR, was developed to address the aforementioned drawbacks. Mostly developed for research centers and academia, SFCW boasts a high dynamic range and low cost of system manufacturing, while producing comparable if not identical results to the impulse counterpart. However, data procurement speed is an inherent problem in SFCW GPR, which seems to keep impulse radar in the lead for production and development. I am proposing a novel approach to elevate SFCW's data acquisition speed and its scanning efficiency altogether. This approach combines an encoding method called orthogonal frequency division multiplexing (OFDM) and an emerging paradigm called compressive sensing (CS). In OFDM, a digital data stream, the transmit signal, is encoded on multiple carrier frequencies. These frequencies are combined in such a way to achieve orthogonality between the carrier frequencies, while mitigating any interference between said frequencies. In CS, a signal can be potentially reconstructed from a few samples below the standardized Nyquist rate. A novel design of the SFCW GPR architecture coupled with the OFDM-CS algorithm is proposed and evaluated using ideal channels and realistically modelled bridge decks.