Title :
On the capacity penalty due to input-bandwidth restrictions with an application to rate-limited binary signaling
Author :
Shamai, Shlomo ; Bar-David, I.
Author_Institution :
Dept. of Electr. Eng., Technion, Haifa, Israel
fDate :
5/1/1990 12:00:00 AM
Abstract :
It is well known that in cases of unlimited channel bandwidth and unconstrained signal spectrum, a constant-envelope constraint on the signal does not reduce the capacity of an additive Gaussian channel, as compared to its capacity with the optimum signals of similar average power. For band-limited channels of bandwidth B, however, the constant-envelope constraints does reduce the capacity, even if the signals themselves are allowed arbitrarily rapid alternations (transitions). Here, using a recently introduced technique, it is shown that the capacity is further drastically reduced if, in addition to the channel limitation, the constant-envelope signal is itself restricted to a small fractional out-of-band power ε. Ideal lowpass and bandpass channel filters are considered, and upper bounds on the capacity are derived. The out-of-band power restrictions can be met, for example, by restricting the average rate of transitions ρ of the binary input signal. It is shown that in this case, ε⩽0.573 ρ/B, which is a result used to upper-bound the capacity for binary signaling under an average-transition-rate constraint
Keywords :
band-pass filters; channel capacity; filtering and prediction theory; information theory; low-pass filters; signalling (telecommunication networks); white noise; AWGN; additive Gaussian channel; average-transition-rate constraint; band-limited channels; bandpass channel filters; binary input signal; channel capacity; constant-envelope constraints; constant-envelope signal; fractional out-of-band power; input-bandwidth restrictions; lowpass channel filters; rate-limited binary signaling; upper bounds; Additive white noise; Band pass filters; Gaussian channels; Upper bound;
Journal_Title :
Information Theory, IEEE Transactions on