Practical Issues inImplementing Analog-to-Inform ation Converters

Thestability andprogrammability ofdigital sig- nalprocessing systems hasmotivated engineers tomovethe analog-to-digital conversion (ADC)process closer andcloser to thefront endofmanysignal processing systems inorderto perform asmuchprocessing aspossible inthedigital domain. Unfortunately, manyimportant applications, including radar and communication systems, involve wideband signals thatseriously stress modern ADCs;sampling these signals abovetheNyquist rate isinsomecases challenging andinothers impossible. While wideband signals bydefinition havealarge bandwidth, often the amount ofinformation they carry persecond ismuchlower; that is, theyarecompressible insomesense. Thefirst contribution of this paper isanewframework forwideband signal acquisition purpose-built forcompressible signals thatenables sub-Nyquist dataacquisition viaananalog-to-information converter (AIC). Theframework isbasedontherecently developed theory of compressive sensitng inwhichasmall numberofnon-adaptive, randomized measurements aresufficient toreconstruct com- pressible signals. Thesecond contribution ofthispaperisan AICimplementation design andstudy ofthetradeoffs andnon- idealities introduced byrealhardware. Thegoalistoidentify andoptimize theparameters thatdominate theoverall system performance.