Over-the-air computation (AirComp) is a promising scheme that exploits the waveform superposition property to enable simultaneous data collection and computation. We propose a scheme involving the encoding of the repetitive transmission of the same symbol at each transmitter over multiple time slots and the design of constellation diagrams, with the aim of minimizing computational errors. We frame this enhancement as an optimization problem. To manage the computational complexity of the optimization, we divide it into two tractable subproblems and use alternating minimization algorithm to solve it. Given the exponential increase in complexity of my problem with respect to its parameters, it becomes essential to validate the performance of my proposed scheme in scenarios involving a larger number of users. Traditional computational resources are insufficient to handle the complexity demands posed by such large-scale simulations. Therefore, I require access to GPU resources to efficiently compute and analyze the performance of my proposed framework. The validation result will be presented as a part in a technical journal after I finish it.