Show that the channel capacity of an ideal AWGN channel with infinite bandwidth is given by C∞ = (1/ln 2)·(S/η) ≈ 1·44 (S/η) bit/sec.
Given that :
S is the average signal power and η/2 is the power spectral density of white Gaussian noise.
10 marks · directive: Show
The core of a good answer
Starting from the Shannon Hartley law C = B log₂(1 + S/(ηB)) and letting B → ∞, using lim (1 + x)^{1/x} = e as x → 0, the capacity approaches (S/η) log₂ e = (1/ln 2)(S/η) ≈ 1.44 S/η bits per second.
An outline written by upsc.help, not an official UPSC answer; every fact checked by a separate review. How we work
What the question asks
The derivation tests the Shannon Hartley theorem and the behaviour of capacity as bandwidth grows: noise power grows with bandwidth, so capacity does not grow without limit but saturates. The method is a limit evaluation with the substitution x = S/(ηB).
Answer outline
Introduction
For an AWGN channel of bandwidth B with two sided noise PSD η/2, the noise power in the band is N = (η/2) × 2B = ηB. The Shannon Hartley theorem gives C = B log₂(1 + S/N).
Step 1: Capacity at bandwidth B
- N = ηB
- C = B log₂(1 + S/(ηB)) bits/s
Step 2: Rearrange
- C = (S/η) × (ηB/S) log₂(1 + S/(ηB))
- Let x = S/(ηB); then C = (S/η) log₂(1 + x)^{1/x}
Step 3: Take the limit B → ∞
- As B → ∞, x → 0
- lim (1 + x)^{1/x} = e as x → 0
- C∞ = (S/η) log₂ e
Step 4: Final form
- log₂ e = 1/ln 2 ≈ 1.4427
- C∞ = (1/ln 2)(S/η) ≈ 1.44 S/η bits/s
Step 5: Interpretation (Shannon limit)
- At capacity, S = E_b C, so C∞ = 1.44 E_b C/η gives E_b/η = ln 2 = 0.693, i.e. −1.6 dB
- No system can communicate reliably with E_b/N₀ below −1.6 dB, however large the bandwidth
Conclusion
Hence C∞ = (1/ln 2)(S/η) ≈ 1.44 S/η bits/s. The result shows that increasing bandwidth alone gives diminishing returns, and that capacity is ultimately limited by the ratio of signal power to noise density.
Examples and data to cite
- The often quoted limit of −1.59 dB for E_b/N₀ applies only to the theoretical case of infinite bandwidth; for finite bandwidth the Shannon limit is always higher.
- Shannon's noisy channel coding theorem dates from 1948 and his paper Communication in the Presence of Noise from 1949; the band limited AWGN capacity formula is known as the Shannon Hartley theorem.
Facts to remember
- Shannon Hartley theorem: C = B log₂(1 + S/N) bits/s for an AWGN channel.
- Noise power in bandwidth B with two sided PSD η/2: N = ηB.
- lim (1 + x)^{1/x} = e as x → 0; log₂ e = 1/ln 2 = 1.4427.
- Shannon limit: E_b/N₀ ≥ ln 2 ≈ 0.693 (−1.59 dB).
- Capacity is a monotonically increasing but bounded function of bandwidth.