Research On Low-Power And Crosstalk Avoidance Codings For Buses

Due to the large load capacitance and increasingly serious inter-wire coupling, deep submicron buses are facing many problems like power, delay and reliability. Low-power bus codings can significantly reduce the dynamic power consumed on the bus by decreasing the self transitions or coupling transitions of bus data. Bus crosstalk avoidance codings can improve bus speed and alleviate crosstalk effects by eliminating the worst-case crosstalk patterns.This paper analyzes the relationships between energy consumption, crosstalk delay and data transition patterns for deep submicron buses, and presents typical low-power bus coding methods and crosstalk avoidance coding methods. Low-power bus coding methods can be divided into methods applied to address buses and ones applied to data buses. Crosstalk avoidance coding methods have two encoding strategies, i.e. spatial encodings and temporal encodings. On the basis of summarizing the research results of former people, this paper proposes a novel low-power encoding method to reduce the power consumption of the data output interface of CMOS image sensors, and also an adaptive spatio-temporal encoding method to eliminate the worst-case crosstalk.The above low-power encoding which freezes the MSBs of the interface as the MSBs of current pixel signal are equal to that of previous one of the same color can reduce the power of the interface by 24.2%, with small cost of codec circuits and no redundant wire. The adaptive spatio-temporal encoding combines the advantages of spatial encodings and temporal encodings. Its extra added lines are much less than spatial encodings, and data transmission efficiency is much higher than temporal encodings. Moreover, it is more energy efficient than spatial encodings and temporal encodings.