Skip to main content

Bit error probability for binary ASK

Short answer

Assuming coherent detection, AWGN, equally likely symbols, and known carrier phase. The bit error probability for binary ASK is P_e = Q(√(E_b/N_0)), where E_b = A^2 T_b/4 is the average energy per bit. This requires 3 dB more energy than coherent BPSK.

With necessary assumption, derive the expression for bit error probability for binary ASK system.

Binary ASK (on-off keying) sends a carrier burst for "1" and nothing for "0". With coherent (matched-filter/correlator) detection in AWGN, the bit error probability is Pe=Q(Eb/N0)P_e = Q\left(\sqrt{E_b/N_0}\right), where EbE_b is the average energy per bit.

Assumptions

  1. Signals over 0≤t≤Tb0 \le t \le T_b:
s1(t)=Acos⁡ωct  ("1"),s0(t)=0  ("0")s_1(t) = A\cos\omega_c t \ \ (\text{"1"}), \qquad s_0(t) = 0 \ \ (\text{"0"})

with fcf_c an integer multiple of 1/Tb1/T_b. 2. Channel adds white Gaussian noise n(t)n(t) of two-sided PSD N0/2N_0/2. 3. Symbols are equally likely; carrier phase and bit timing are known (coherent detection).

Receiver

 r(t) --->(X)--->[ integrate 0..Tb ]--> y --> [ y > Vth ? ] --> 1/0
           ^                                   sample at Tb
     cos(wc t)  (local coherent carrier)

Correlator output

y=∫0Tbr(t)cos⁡ωct dty = \int_0^{T_b} r(t)\cos\omega_c t\,dt
  • For "1": signal part =∫0TbAcos⁡2ωct dt=ATb2= \int_0^{T_b}A\cos^2\omega_c t\,dt = \dfrac{AT_b}{2}
  • For "0": signal part =0= 0

Noise at the output: no=∫0Tbn(t)cos⁡ωct dtn_o = \int_0^{T_b} n(t)\cos\omega_c t\,dt is Gaussian, zero mean, with variance

σ2=N02∫0Tbcos⁡2ωct dt=N0Tb4\sigma^2 = \frac{N_0}{2}\int_0^{T_b}\cos^2\omega_c t\,dt = \frac{N_0T_b}{4}

Threshold. For equal priors and equal variances, the optimum threshold is midway: Vth=ATb/4V_{th} = AT_b/4.

Error probabilities

P(e∣0)=P(no>ATb4)=Q ⁣(ATb/4σ)P(e∣1)=P(no<−ATb4)=Q ⁣(ATb/4σ)\begin{aligned} P(e|0) &= P\left(n_o > \tfrac{AT_b}{4}\right) = Q\!\left(\frac{AT_b/4}{\sigma}\right) \\ P(e|1) &= P\left(n_o < -\tfrac{AT_b}{4}\right) = Q\!\left(\frac{AT_b/4}{\sigma}\right) \end{aligned} Pe=12P(e∣0)+12P(e∣1)=Q ⁣(ATb/4N0Tb/4)=Q ⁣(A2Tb4N0)P_e = \tfrac12 P(e|0) + \tfrac12 P(e|1) = Q\!\left(\frac{AT_b/4}{\sqrt{N_0T_b/4}}\right) = Q\!\left(\sqrt{\frac{A^2T_b}{4N_0}}\right)

In terms of energy. Energy of the "1" pulse E1=A2Tb/2E_1 = A^2T_b/2; the "0" has zero energy, so average energy per bit Eb=A2Tb/4E_b = A^2T_b/4. Hence

Pe=Q ⁣(EbN0)=Q ⁣(E12N0)=12 erfc ⁣(Eb2N0)P_e = Q\!\left(\sqrt{\frac{E_b}{N_0}}\right) = Q\!\left(\sqrt{\frac{E_1}{2N_0}}\right) = \frac12\,\text{erfc}\!\left(\sqrt{\frac{E_b}{2N_0}}\right)

Remarks

  • Coherent BPSK gives Q(2Eb/N0)Q(\sqrt{2E_b/N_0}), so ASK needs 3 dB more average energy for the same PeP_e.
  • Non-coherent (envelope) ASK gives approximately Pe≈12e−Eb/2N0P_e \approx \tfrac12 e^{-E_b/2N_0}, slightly worse but simpler.

✓ Answer checked by Claude Opus 5.5 (high effort)

All Communication Systems chapter 9 questions →