Speaker
Description
CMOS pixel sensors have demonstrated high detection efficiency, high resolution, low material budgets, low costs, and potentially good radiation tolerance, high speed and low power consumption for the particle tracking in high energy physics experiments. In this paper, we report a study of some new readout strategies that reducing the signal readout times of CPS for the possibility of increasing the readout speed and reducing the power consumption. Firstly, owing to the sparsity of the particle imaging in high energy physics experiments, Compressive Sensing theory is applied. Considering the hardware implementation, the observation matrix is applied for the signals in a column and the column signals are parallel read out. Observation matrix similar with Bernoulli matrix and Orthogonal Matching Pursuit algorithm are used for the sampling and the image reconstruction, respectively. The simulation results indicate the particle images can be well reconstructed with less than 100 times of samplings for either analogue pixels or digital pixels in a matrix of 1024×1024 during 100 frames of various input particle images. The maximal absolute error is in order of $10^{-12}$. Secondly, we are enlightened by the idea of Compressive Sensing and propose to read out the digital pixel signals with a binary search algorithm. The simulation results indicate all the nonzero data can be found in less than 100 times of readouts for various input particle images of 1024×1024 pixels. In one word, the readout strategies of both compressive sensing and binary search algorithm can reduce the readout times sharply comparing with the mature rolling-shutter readout. The CPS chips with these new readout strategies are under design. The readout speed and the power consumption will be evaluated in the future. In addition, the readout times could be reduced in advance by applying an observation matrix for the whole pixel matrix, then the hardware chip design will be the crucial problem.
