Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


ucin1195585950.pdf (3.47 MB)

ETD Abstract Container

Abstract Header

SPLINE CENTER AND RANGE REGRESSION TECHNIQUE AND ITS APPLICATION TO VARIATION AWARE PERFORMANCE MACROMODELING OF ANALOG CIRCUITS

KOMMINENI, BALAJI
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1195585950

Abstract Details

2007, MS, University of Cincinnati, Engineering : Electrical Engineering.
Regression is an essential part of engineering methodologies. It plays a critical role in modeling unknown relationships among variables. However when measurement accuracy becomes very important simple regression techniques fail to capture these relationships. This is due to the fact that measured data is prone to much variation due to noise which is unavoidable. Thus, a more robust regression technique is needed which can be used even in the presence of noise. This work is aimed at the development of such a regression technique. The technique is applied to the problem of analog circuit synthesis for generating process variation tolerant analog designs. Noisy data can be modeled as intervals. Interval data types represent a range of values which a variable can assume within a lower bound and an upper bound. Thus, the problem translates to modeling interval valued data. But, the regular real regression techniques cannot be extended to intervals due to the inherent problems of interval arithmetic. Performing real valued arithmetic on interval data and a regression technique which allows such a manipulation form a possible solution. Several such techniques have been proposed in literature which make use of the center and range information of the intervals. Spline Center and Range Regression is a new technique proposed in this work. The Spline Center And Range Regression technique is capable of interpolating the lower and upper bounds of interval valued data based on the center and range information of the intervals. The Spline Center and Range Regression approach interpolates on the center and range values of an interval independently which are then combined to obtain the interval bound. For interpolating the center and the range it uses spline regression which is a type of regression capable of working on highly non-linear data and at the same time provides excellent accuracy. Thus, combining splines with the above mentioned approach gives a highly accurate interval regression technique as compared to it’s linear counterparts. The Spline Center and Range Regression technique is applied to the VLSI design to tackle the effects of process variations on analog circuits. Process variations cause the performance parameters used for designing the circuit to vary from the intended value once the circuit is manufactured. Performance macromodels have been used extensively for the analog circuit sizing process. Conventional macromodels have failed to take into account the effects of process variations. To overcome this problem new variation aware performance macromodels are developed using SCRR. These macromodels are then used to synthesize process variation tolerant analog circuits.
Dr. Ranga Vemuri (Advisor)
80 p.

Recommended Citations

Citations

  • KOMMINENI, B. (2007). SPLINE CENTER AND RANGE REGRESSION TECHNIQUE AND ITS APPLICATION TO VARIATION AWARE PERFORMANCE MACROMODELING OF ANALOG CIRCUITS [Master's thesis, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1195585950

    APA Style (7th edition)

  • KOMMINENI, BALAJI. SPLINE CENTER AND RANGE REGRESSION TECHNIQUE AND ITS APPLICATION TO VARIATION AWARE PERFORMANCE MACROMODELING OF ANALOG CIRCUITS. 2007. University of Cincinnati, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1195585950.

    MLA Style (8th edition)

  • KOMMINENI, BALAJI. "SPLINE CENTER AND RANGE REGRESSION TECHNIQUE AND ITS APPLICATION TO VARIATION AWARE PERFORMANCE MACROMODELING OF ANALOG CIRCUITS." Master's thesis, University of Cincinnati, 2007. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1195585950

    Chicago Manual of Style (17th edition)

ucin1195585950
716
© 2007, all rights reserved.
This open access ETD is published by University of Cincinnati and OhioLINK.