Abstract: Existing full-pulse structured waveform and processing methods, such as phase-coded waveform matched filtering, have inherent defects of poor Doppler tolerance Additionally, linear frequency modulation (LFM) waveforms with windowing processing degrade range resolution and signal-to-noise ratio (SNR) gain, making them insufficient for high-speed multi-target detection requirements. To address these limitations, this paper proposes a multi-subpulse structured waveform design and processing method for high-speed target detection. First, an echo model of the multi-subpulse structured waveform is constructed. The range-Doppler response function is derived using segmented subpulse compression processing and inter-subpulse coherent integration. Next, based on the target range-velocity region of interest, an optimization problem for the multi-subpulse structured waveform is formulated under a constant modulus constraint, with the objective of minimizing the weighted integrated range-Doppler sidelobe level. Finally, a coordinate descent (CD) optimization framework is introduced to decompose the high-dimensional non-convex-constrained optimization problem into iterative solutions of multiple one-dimensional subproblems, for which closed-form solutions are derived. Simulation results demonstrate that the proposed multi-subpulse structured waveform exhibits superior Doppler tolerance and lower local range-Doppler sidelobe levels compared to the LFM waveforms, ambiguity-function-optimized waveforms, and LFM-noise waveforms. It further achieves enhanced high-speed target detection capability in cognitive multi-target scenarios.