This paper presents a level-crossing sampling based spike detection and feature extraction algorithm to reduce data rate and power consumption in neural recording systems. I have the following questions regarding the presentation of the current manuscript.
1. What is the difference between a. and a. Motivation: In the introduction, the authors mentioned that the power consumption in wireless communication is the primary consideration in the system. However, the receiver side, including the spike detection and feature extraction, are usually not power-limited. The authors should explain what is the power target for the overall implanted blocks and what is the power budget for the receiver side. 2. What is the difference between…show more content… The authors should quantitatively compare the performance of their proposed system to these systems to show the advantage of the proposed system. 3. What is the difference between a'smart' and a'smart'? Proposed method: Although LCS looks simple, as shown in Fig.2, it cannot detect the spike peaks for the waves since the gap between each sampling voltage level is high, and the quantization accuracy is pretty low. If the gap voltage is reduced for a higher accuracy, LCS may generate a higher data rate than Nyquist Rate ADC. The authors should explain how this affects the processing algorithm. 4. What is the difference between a.. Proposed algorithm: As discussed above, LCS system can not detect signal peaks, non of the peaks in Fig. 2 is detected by LCS. So the first step of the proposed method (peak detection) using UD changes is not solid. Due to the large gap between the reference voltages, the duration of the peak estimated from LCS may also be not accurate enough. 5. What is the difference between a'smart' and a'smart'? Hardware implementation and comparison. This work has not been implemented in the hardware, but only with software synthesis. The
