Fig. 1
Randomness of Biological Physically Unclonable Function (Bio-PUF). a Schematic of process flow for generating a Bio-PUF. First, the T cells are cultured and suspended in individual well(s) and then transferred to an imaging system with an onstage incubator. These T cells are subsequently allowed to form colonies for several hours (~20 h). Finally, the colonized T cell population is imaged and processed using a computer for generating the Bio-PUF. b False color optical image of a randomly chosen post-colonization T cell population with precise cell identification (purple). c Construction of Bio-PUF with 64 × 64 pixels (bits). Pixels occupied by T cells are assigned digital “1” (white), whereas, unoccupied pixels are assigned digital “0” (black). d Entropy (Ex,y) content of the Bio-PUF along both axes. e 64 1D keys each with 64 bits associated the Bio-PUF. f Probability mass function (PMF) of the intra-Hamming distance (Hintra) between the 64C2 = 2016 pairs of 1D keys. The mean Hintra extracted using a binomial fit was found to be ≈29, which suggests close to maximum uniqueness for the 1D keys and hence the Bio-PUF. g Histogram of correlation coefficients (CCintra) between the 1D keys. The mean CCintra extracted using a Gaussian fit (inset) was found to be 0.01, which confirms lack of any correlation between the 1D keys. h Autocorrelation coefficient (ACF) for an arbitrarily chosen 1D key as a function of the lag. i 3D plot of the ACF for all 64 keys. The presence of low amplitude and narrow spikes indicate none to minimal memory effect in colonized T cell population and rationalizes the strength of Bio-PUF
