# UA MATH636 信息论7 高斯信道

### UA MATH636 信息论7 高斯信道

Y t = ( X t + Z t ) ∗ h t Y_t = (X_t + Z_t)*h_t

H ( f ) = I ( ∣ f ∣ ≤ W ) H(f)=I(|f| \le W)

Y t = ( X t + Z t ) ∗ h t = X t ∗ h t + Z t ∗ h t Y_t = (X_t + Z_t)*h_t = X_t *h_t + Z_t*h_t

# 高斯信道的容量

P S D ( f ) = N 0 2 PSD(f)=\frac{N_0}{2}

R Z ∗ h ( f ) = N 0 2 I ( ∣ f ∣ ≤ W ) R_{Z*h}(f) = \frac{N_0}{2}I(|f| \le W)

1 2 ln ⁡ ( 1 + P N ) \frac{1}{2}\ln \left(1 + \frac{P}{N} \right)

A C F ( τ ) = F − 1 ( P S D ) = N 0 2 s i n c ( τ ) ACF(\tau) = \mathcal{F}^{-1}(PSD) = \frac{N_0}{2} sinc(\tau)

A C F s a m p l e ( n 2 W ) = N 0 2 I ( n = 0 ) , n ∈ N ACF_{sample}(\frac{n}{2W}) = \frac{N_0}{2}I(n = 0),n\in \mathbb{N}

E [ Z ~ ( i 2 W ) Z ~ ( i 2 W ) ] = N 0 2 E [ Z ~ ( i 2 W ) Z ~ ( j 2 W ) ] = 0 E[\tilde{Z}(\frac{i}{2W})\tilde{Z}(\frac{i}{2W})]=\frac{N_0}{2} \\ E[\tilde{Z}(\frac{i}{2W})\tilde{Z}(\frac{j}{2W})]=0

1 2 ln ⁡ ( 1 + P N ) = 1 2 ln ⁡ ( 1 + P 2 W N 0 2 ) \frac{1}{2}\ln \left(1 + \frac{P}{N} \right) = \frac{1}{2}\ln \left(1 + \frac{\frac{P}{2W}}{\frac{N_0}{2}} \right)

C = W ln ⁡ ( 1 + P N 0 W ) C = W \ln \left(1 + \frac{P}{N_0W} \right)

C ∞ = lim ⁡ W → ∞ C = lim ⁡ W → ∞ W ln ⁡ ( 1 + P N 0 W ) = P N 0 C_{\infty}=\lim_{W \to \infty} C = \lim_{W \to \infty} W \ln \left(1 + \frac{P}{N_0W} \right) = \frac{P}{N_0}

C ∞ = P N 0 ln ⁡ 2 C_{\infty} = \frac{P}{N_0 \ln 2}

## Shannon Limit

R b = k T R_b = \frac{k}{T}

E = P T = k E b E = PT = kE_b
E b E_b 的含义是energy per bit，则
E b = P T k E_b = \frac{PT}{k}

R b ≤ C ∞ ⇔ C ∞ R b > 1 ⇒ P / N 0 ln ⁡ 2 T k = E b N 0 ln ⁡ 2 > 1 ⇒ E b N 0 > ln ⁡ 2 R_b \le C_{\infty} \Leftrightarrow \frac{C_{\infty}}{R_b}>1 \\ \Rightarrow \frac{P/N_0}{\ln 2} \frac{T}{k} = \frac{E_b}{N_0 \ln2} >1 \Rightarrow \frac{E_b}{N_0} > \ln 2

## Beyond Shannon Limit

R b = k T = k N 2 W = 2 W k N = 2 W R R_b = \frac{k}{T} = \frac{k}{\frac{N}{2W}} = 2W \frac{k}{N} = 2W R

R b = 2 W R < W ln ⁡ ( 1 + P N 0 W ) R_b = 2WR < W \ln \left(1 + \frac{P}{N_0W} \right)

P N 0 W = k E b T N 0 W = 2 R E b N 0 \frac{P}{N_0W} = \frac{kE_b}{TN_0W} = \frac{2RE_b}{N_0}

2 R < ln ⁡ ( 1 + 2 R E b N 0 ) 2R < \ln (1+\frac{2RE_b}{N_0})

E b N 0 > 2 2 R − 1 2 R \frac{E_b}{N_0} > \frac{2^{2R}-1}{2R}
R → 0 R \to 0 时，这个下界趋近于Shannon Limit。

• 点赞
• 评论
• 分享
x

海报分享

扫一扫，分享海报

• 收藏
• 打赏

打赏

一个不愿透露姓名的孩子

你的鼓励将是我创作的最大动力

C币 余额
2C币 4C币 6C币 10C币 20C币 50C币
• 举报
• 一键三连

点赞Mark关注该博主, 随时了解TA的最新博文

08-06 9046

04-19 4840
02-23 4705
06-16 272
11-06 4724
09-19 3207
12-01 4525
08-15 1万+
07-20 79
09-09 3689
12-18 1万+
03-18 3561
01-28 6443
08-05 1万+
08-05 1万+
09-20 7万+
04-21 1255