This poster presents a structured model for industry-partnered senior capstone education in Computer Engineering (CE) at University of California, Santa Barbara. Student teams engage with industry sponsors---spanning aerospace, defense, IoT, ML, and embedded systems---through a pipeline covering project solicitation, elevator pitches, sponsor networking, and preference-based team formation. Now in its tenth year, the program has engaged 35 unique sponsors (30 from industry), serving approximately 50-60 students per year across 11-15 projects. We distill ten years of operational experience into a compact program model, highlighting project selection, faculty evaluation, and concrete success stories.

In this paper, we analyze the performance of transmit beamforming on multiple-antenna Rayleigh fading channels with imperfect channel feedback. We characterize the feedback imperfections in terms of noisy channel estimation, feedback delay, and finite-rate channel quantization. We develop a general framework, that is valid for an arbitrary two-dimensional linear modulation, to capture the aforementioned imperfections and derive the symbol and bit error probability expressions for both M-PSK and M-ary rectangular QAM constellations with Gray code mapping. We show that the proposed analytical formulation is valid for a frequency-domain duplexing system with/without finite-rate channel quantization and a time-domain duplexing system. We validate the accuracy of the analysis through simulations, and assess the relative effects of channel estimation inaccuracy, feedback delay, and finite-rate quantization on the symbol and bit error performances for various constellations.

We address the problem of analyzing the effect of channel estimation errors on the average block error probability (BLEP) of transmit beamforming multiple input single output (MISO) system in a slowly varying Rayleigh fading wireless channel. For this purpose, we develop an accurate model of estimation errors in a block fading context and derive an analytical expression quantifying the impact of channel estimation errors on feedback MISO systems. In particular, the block fading model implies that the channel estimate and its error component are fixed for the entire block and an appropriate performance criteria is the average BLEP, a more complex metric to analyze. The derived closed form analytical expression for the average BLEP is validated by simulations.

In this paper, the problem of finite-rate feedback for spatially and temporally correlated Rayleigh fading multiple input single output (MISO) channels with estimation errors at the receiver and feedback delay is addressed. A model that captures estimation errors, feedback delay, and finite-rate quantization of the channel is developed. A novel codebook design algorithm that minimizes the loss in ergodic capacity is proposed. Simulation results show that the new codebook designed under the consideration of estimation errors and feedback delay outperforms the codebook designed assuming ideal conditions. Analysis for the loss in ergodic capacity for spatially i.i.d channels with channel estimation errors and delay (EED) is presented and validated through simulations.

In this paper, utilizing high resolution quantization theory, we analyze the loss in average symbol error probability (SEP) for finite rate feedback MISO systems with rectangular M-QAM constellation. Assuming perfect channel estimation, no-feedback delay and error-less feedback, for spatially i.i.d and correlated channels we derive analytical expressions for loss in average SEP due to finite-rate channel quantization. We then consider the high-SNR regime and show that the loss associated with correlated case is related to the loss associated with the i.i.d case by a scaling constant given by the determinant of the correlation matrix. We also present simulation results in support of the analytical expressions.

We analyze the performance of transmit beamforming on multiple-antenna Rayleigh fading channels with imperfect channel feedback. The feedback imperfections are characterized by noisy channel estimation, feedback delay, and finite rate channel quantization. We develop a general framework, valid for an arbitrary two-dimensional linear modulation, that captures the aforementioned imperfections, and derive the average symbol and bit error probability expressions for M- PSK constellation. We show that the proposed analytical formulation is valid for frequency-domain duplexing system with/without finite rate channel quantization and time-domain duplexing system. The simulation results show accurate agreement with the analytical expressions.