A Study of Contact Network Generation for Cyber-bullying Detection

A Study of Contact Network Generation for Cyber-bullying Detection

The two observable differences here are the lower magnitudes and the difference in the shape of the envelope formed by the secondary noise tones. Since the shape of this envelope is caused by the resonant ringing due to system parasitics, this implies that the shape of the envelope should somewhat flatten and shift slightly higher frequencies for the COB measurements. a closer look at the ringing characteristics in the time domain. The waveforms seen here exhibit very similar characteristics with the noise waveforms for the high inductance package in As before, there is a smaller higher frequency ringing due to the capacitive coupling component and larger spikes due to the switching noise from the CBL PRNG. These components can be easily seen in the middle plot when the clock frequency is reduced to For the NCL plot on the bottom, the waveform actually looks very different than the one from In this case, there seems to be a much higher frequency ringing than before.

This observation appears to contradict our previous argument regarding the shape of the waveform since the operating speed of the NCL PRNG should not have changed so drastically when the package parasitics are reduced.The explanation for this difference lies in the assumption made above where the offset in switching times is a small fraction of the resonant period. This resonant period is equivalent to the resolution of the switching transitions seen in the time domain. A lower inductance package means that finer and more abrupt changes in the switching distribution will be visible in the substrate noise. Referring back to it can be seen that right after each switching peak, there is a slight dip in the number of transitions followed by a secondary peak.

A Study of Contact Network Generation for Cyber-bullying Detections

These secondary peaks are the sources of the additional ringing seen in the NCL PRNG time domain waveform of Comparison of the RMS value of the substrate noise in heavily and lightly doped substrates top synchronous PRNG, middle synchronous PRNG with a clock, and bottom asynchronous PRNG. C. Lightly Doped Substrate A technique often employed to reduce substrate noise coupling for SOC applications is a lightly doped process. The increased resistivity for the substrate in this process compared to the traditional heavily-doped substrate serves to attenuate and/or isolate the coupling to the analog circuits. Whereas a heavily doped substrate can be modeled with a single node approximation, physical separation becomes more of a factor for lightly doped substrates.

In order to quantify the improvement of the substrate noise of CBL and NCL circuits for this process, the equivalent PRNG chip was also fabricated in a lightly-doped process. a summary of the measured RMS value of the substrate noise for both the high inductance and low inductance cases for the two processes. The top plot compares the substrate noise for the heavily and lightly doped processes when the CBL PRNGs are clocked at the same equivalent operating speed as their respective NCL PRNGs. Note that for this process, the equivalent operating speed for the lightly doped NCL PRNG is As seen in the figure, the measured substrate noise in the high inductance package is slightly lower for the lightly doped chip. When both of the CBL PRNGswere clocked at a slightly larger improvement was seen. In the low inductance package, however, there is no noticeable improvement between the coupled noise in the two different substrates. This same trend was also found to be true for the NCL PRNGs.