ABSTRACT
The ultimate data rate of any communication channel is constrained by the bandwidth and the signal-
to-noise ratio (SNR) of the channel [1]. As a result, the quantity of noise at any point in a signal chain
places a direct limit on how many bits per second can pass through that point for a given bandwidth and
power level. In a typical system such as a wireless receiver, noise can be introduced in a variety of ways,
including by the active devices (both bipolar and MOS) used to implement the circuits. Noise may be
added directly in band with the signal, such as in the case of a low-noise amplifier (LNA) at the front end
of a receiver chain or in the baseband amplifier near the end of the chain. In addition, nonlinearities may
allow noise to enter the channel band indirectly. For example, transistor noise in a voltage-controlled
oscillator (VCO) circuit introduces phase noise in the output sinusoid. When serving as a local
oscillator, such a signal passed to a mixer along with the data signal can cause out-of-band (e.g.,
adjacent channel) energy to fold in-band, contributing interference that degrades channel capacity in
a manner analogous to noise. While the effects of noise can be countered by increasing the energy per
bit, which translates into signal power for a given bit rate, many applications, particularly portable
systems such as cellular phones, GPS receivers, and WLAN-capable PDAs, are constrained in transmitted
or received signal power by battery life limits, transmitter-receiver distance, feasible antenna and
package size or standards and regulatory requirements. Supply voltage and amplifier gain compression
can impose an upper signal power limit as well. Thus, communication system designers seeking an
attractive trade-off between throughput, signal power, and bit-error rate demand technologies capable of
processing high-frequency signals while introducing as little noise as possible.
