Leakage compensation in analog random modulation pre-integration architectures for biosignal acquisition

As Compressed Sensing (CS) emerges as an innovative approach for analog-to-information conversion, more realistic models for studying and coping with the non-idealities of real circuits are required. In this paper we consider the effect of the voltage drop due to leakage currents in the random modulation pre-integration approach, which is the most common CS architecture. In particular we focus on switched capacitor implementations, and we show that leakage currents may significantly alter the acquired information especially when integration time is long as it happens, for example, in biosensing applications. With a simple but realistic circuit model we show that the voltage drop has two contributions. The first is signal independent and causes an offset in the measurement, while the second is signal dependent. To cope with these effects we propose two compensation techniques that ensure signal reconstruction even in the presence of measurement degradation due to leakage.