Abstract—According to the development of information technologies, a high-speed similar data retrieval system is becoming important to retrieve similar data from mass data in database for character recognition, fingerprint recognition, data compression, color image recognition, and so on. However, the conventional similar data retrieval system implemented by software or hardware using conventional memory is slow, because a computer must compare the called data sequentially. For this reason, in order to retrieve similar data, an associative memory has been studied in recent years. In this paper, by utilizing a neuron CMOS inverter, we propose a novel minimum Manhattan distance retrieving circuit as an important functional block of the associative memory. The proposed circuit is less affected by the influence of the initial charge and the variation of threshold voltage. The effectiveness of the proposed circuit is demonstrated by the Simulation Program with Integrated Circuit Emphasis (SPICE) simulations.
 

Index Terms—Manhattan distance, associative memory, neuron CMOS inverter, time domain

Cite: Yujiro Harada, Kuniaki Fujimoto, Kei Eguchi, Masaaki Fukuhara, and Masahiro Yoshida, "A Minimum Manhattan Distance Retrieving Circuit Using Neuron CMOS Inverters," International Journal of Electronics and Electrical Engineering, Vol. 4, No. 4, pp. 290-295, August 2016. doi: 10.18178/ijeee.4.4.290-295