A new algorithm and parallel architecture for high-speed complex number multiplication is presented, and a prototype chip based on the proposed approach is designed. By employing redundant binary (RB) arithmetic, an N-bit complex number multiplication is simplified to two RB multiplications (i.e., an addition of N RB partial products), which are responsible for real and imaginary parts, respectively. Also, and efficient RB encoding scheme proposed in this paper enables to generate RB partial products without additional hardware and delay overheads compared with binary partial product generation. The proposed approach leads to a highly parallel architecture with regularity and modularity. As a results, it results in much simpler realization and higher performance than the classical method based on real multipliers and adders. As a test vehicle, a prototype 8-b complex number multiplier core has been fabricated using $0.8mu extrm{m}$ CMOS technology. It contains 11,500 transistors on the area of about $1.05 imes 1.34 textrm{mm}^2$. The functional and speed test results show that it can safely operate with 200 MHz clock at $V_{DD}=2.5 V$, and consumes about 90mW.