108 年 國立中央大學通訊工程學系碩士班甲組《通訊系統》

📄 試題原卷 免費註冊後即可對照原始考卷 PDF免費註冊

第 1 題50 分

國立中央大學 108 學年度碩士班考試入學試題
所別:通訊工程學系碩士班 不分組(一般生)
科目:通訊系統
本科考試禁用計算器
共2頁 第1頁

  1. (50%) For a baseband communication system as shown below, where w(t)w(t) is the additive white Gaussian noise with a power spectral density N0/2N_0/2 and p(t)/h(t)p(t)/h(t) denote the impulse responses of TX/RX filters, (a) plot the Fourier spectrum As(f)=F{as(t)}A_s(f) = \mathcal{F}\{a_s(t)\} in the range ∣f∣≤2fs|f|\le 2f_s when A(f)=F{a(t)}={1,∣f∣≤0.4fs0,otherwiseA(f) = \mathcal{F}\{a(t)\} = \begin{cases} 1, & |f|\le 0.4f_s \\ 0, & \text{otherwise} \end{cases} (5%); (b) find the values fpassf_{pass} and fstopf_{stop} such that s(t)=K⋅a(t)s(t) = K \cdot a(t) and fstop−fpassf_{stop} - f_{pass} is maximized for the signal a(t)a(t) given in (a) (5%); (c) plot the waveform r(t)r(t) in the range 0≤t≤5⋅Ts0 \le t \le 5 \cdot T_s when p(t)={2,0≤t≤0.5Ts0,otherwisep(t) = \begin{cases} 2, & 0 \le t \le 0.5T_s \\ 0, & \text{otherwise} \end{cases}, h(t)=p(t)h(t) = p(t), N0=0N_0 = 0 and {a(n⋅Ts)∣n=0∼4}={1,−1,−3,3,−1}\{a(n \cdot T_s) | n=0 \sim 4\} = \{1, -1, -3, 3, -1\} (5%); (d) find the received signal power E{r2(t)}\mathbb{E}\{r^2(t)\} in terms of N0N_0 when h(t)={2,0≤t≤0.5Ts0,otherwiseh(t) = \begin{cases} 2, & 0 \le t \le 0.5T_s \\ 0, & \text{otherwise} \end{cases} and s(t)=0s(t) = 0 (5%); (e) find the values A0,t0A_0, t_0 and E{nk2}\mathbb{E}\{n_k^2\} in r[k]=A0a(k⋅Ts+t0)+nkr[k] = A_0 a(k \cdot T_s + t_0) + n_k such that A0A_0 is maximized when h(t)=p(t)={2,0≤t≤0.5Ts0,otherwiseh(t) = p(t) = \begin{cases} 2, & 0 \le t \le 0.5T_s \\ 0, & \text{otherwise} \end{cases} and N0≠0N_0 \ne 0 (12%); (f) find the optimal decision rule for r[k]r[k] given in (e) when a(k⋅Ts)∈{−3,−1,1,3}a(k \cdot T_s) \in \{-3, -1, 1, 3\} with equal probability (5%); (g) find the decision error probability for the decision rule given in (f) in terms of A0,N0A_0, N_0 and Q(x)Q(x) (8%); (h) find the data rate (bps: bit-per-sec) of the system when a(k⋅Ts)∈{−7,−5,−3,−1,1,3,5,7}a(k \cdot T_s) \in \{-7, -5, -3, -1, 1, 3, 5, 7\} (5%).

Hint: Q(x)=12π∫x∞exp⁡(−y22)dyQ(x) = \frac{1}{\sqrt{2\pi}} \int_{x}^{\infty} \exp(-\frac{y^2}{2}) dy, Pr(n>A)=Q(Aσ)Pr(n > A) = Q(\frac{A}{\sigma}), n∼N(0,σ2)n \sim \mathcal{N}(0, \sigma^2)
Hint: s(t)=a(t)∗p(t)=∑n=−∞∞a(n⋅Ts)p(t−n⋅Ts)s(t) = a(t) * p(t) = \sum_{n=-\infty}^{\infty} a(n \cdot T_s) p(t - n \cdot T_s), F{⋅}\mathcal{F}\{\cdot\}: Fourier Transform

🖼️【此處有附圖,請對照原卷】

🖼️ 本題含圖表,以下為原卷對應頁面:
原卷第 1 頁

登入後即可作答並保存紀錄。

這一題的完整詳解

核心觀念

本題考查取樣脈衝列的頻譜複製、發射與接收濾波器的等效脈衝響應、AWGN 通過線性濾波器後的雜訊變異數,以及等概多階 PAM 的最佳判決。

令 fs=1/Tsf_s=1/T_s。取樣脈衝列的傅立葉轉換會使原頻譜以 fsf_s 為間隔重複;濾波器輸出雜訊的變異數則為

σn2=N02∫−∞∞∣h(t)∣2 dt.\sigma_n^2=\frac{N_0}{2}\int_{-\infty}^{\infty}|h(t)|^2\,dt.

解題方法與推導

(a) 繪出 As(f)A_s(f)

題目中的取樣信號為

as(t)=a(t)∑n=−∞∞δ(t−nTs).a_s(t)=a(t)\sum_{n=-\infty}^{\infty}\delta(t-nT_s).

因此

As(f)=fs∑m=−∞∞A(f−mfs).A_s(f)=f_s\sum_{m=-\infty}^{\infty}A(f-mf_s).

原頻譜在 ∣f∣≤0.4fs|f|\le 0.4f_s 時為 1,所以每個頻譜複本的高度為 fsf_s、寬度為 0.8fs0.8f_s,中心位於 mfsmf_s。在題目指定的範圍 ∣f∣≤2fs|f|\le 2f_s 內:

頻率區間As(f)A_s(f)
[−2,−1.6]fs[-2,-1.6]f_sfsf_s
[−1.4,−0.6]fs[-1.4,-0.6]f_sfsf_s
[−0.4,0.4]fs[-0.4,0.4]f_sfsf_s
[0.6,1.4]fs[0.6,1.4]f_sfsf_s
[1.6,2]fs[1.6,2]f_sfsf_s
其餘區間00

每個區間都是高度 fsf_s 的矩形頻譜;各複本之間有頻譜空隙。

(b) 求 fpassf_{pass} 與 fstopf_{stop}

題目要求 s(t)=Ka(t)s(t)=K a(t),因此在原信號頻帶 ∣f∣≤0.4fs|f|\le 0.4f_s 內,發射濾波器必須保留信號頻譜;又因為取樣造成的第一個相鄰頻譜複本從 ∣f∣=0.6fs|f|=0.6f_s 開始,該處起必須進入阻帶,才能避免複本出現在輸出。

故通帶邊緣至少要涵蓋 0.4fs0.4f_s,而阻帶邊緣需不晚於 0.6fs0.6f_s。取最大轉換帶寬時,兩個邊緣分別位於這兩個界線:

fpass=0.4fs,fstop=0.6fs.f_{pass}=0.4f_s,\qquad f_{stop}=0.6f_s.

在原信號頻帶內,As(f)=fsA_s(f)=f_s,因此通帶增益取 K/fsK/f_s 即可使 S(f)=KA(f)S(f)=K A(f)。

(c) 繪出 r(t)r(t)

發射與接收濾波器相同,且

p(t)=h(t)={2,0≤t≤0.5Ts,0,其他.p(t)=h(t)= \begin{cases} 2,&0\le t\le 0.5T_s,\\ 0,&\text{其他}. \end{cases}

等效脈衝響應為 g(t)=p(t)∗h(t)g(t)=p(t)*h(t)。兩個高度為 2、寬度為 0.5Ts0.5T_s 的矩形脈衝卷積後,得到支撐區間為 [0,Ts][0,T_s] 的三角波:

g(t)={4t,0≤t≤0.5Ts,4(Ts−t),0.5Ts<t≤Ts,0,其他.g(t)= \begin{cases} 4t,&0\le t\le 0.5T_s,\\ 4(T_s-t),&0.5T_s<t\le T_s,\\ 0,&\text{其他}. \end{cases}

其峰值為 g(0.5Ts)=2Tsg(0.5T_s)=2T_s。因此

r(t)=∑n=04a(nTs)g(t−nTs).r(t)=\sum_{n=0}^{4}a(nT_s)g(t-nT_s).

相鄰三角波只在端點相接,端點值皆為零,所以輸出是依序乘上各符號的三角波:

時間r(t)r(t)
0≤t≤Ts0\le t\le T_s在 0.5Ts0.5T_s 達峰值 2Ts2T_s
Ts≤t≤2TsT_s\le t\le 2T_s在 1.5Ts1.5T_s 達峰值 −2Ts-2T_s
2Ts≤t≤3Ts2T_s\le t\le 3T_s在 2.5Ts2.5T_s 達峰值 −6Ts-6T_s
3Ts≤t≤4Ts3T_s\le t\le 4T_s在 3.5Ts3.5T_s 達峰值 6Ts6T_s
4Ts≤t≤5Ts4T_s\le t\le 5T_s在 4.5Ts4.5T_s 達峰值 −2Ts-2T_s

並且 r(nTs)=0r(nT_s)=0,n=0,…,5n=0,\ldots,5。每一段在表列峰值與前後零點間作線性變化。

(d) 接收雜訊功率

🔒

後續完整解題步驟與【答案】

免費註冊,享三天全站完整詳解閱覽。

免費註冊

第 2 題8 分

  1. (8%) Consider the received FM signal x(t)=Acos⁡(2πfct+2πfd∫−∞tm(τ)dτ)+n(t)x(t) = A \cos(2\pi f_c t + 2\pi f_d \int_{-\infty}^{t} m(\tau) d\tau) + n(t), where m(t)m(t) is the lowpass message signal with bandwidth WW and n(t)n(t) is the AWGN with double-sided PSD N0/2N_0/2. For a fixed value of fdf_d, if we increase the value of fsf_s, will the SNR of the demodulated signal increase also? Please explain your answer.

登入後即可作答並保存紀錄。

這一題的完整詳解

核心觀念

題目中的 fsf_s 未明確定義;以下依通訊系統常用記號,將它視為單音調變頻率。FM 解調後的訊號雜訊比,關鍵在於調變指數:頻率偏移固定時,調變頻率愈高,調變指數愈小。

對 m(t)=Amcos⁡(2πfst)m(t)=A_m\cos(2\pi f_s t),峰值頻率偏移為 Δf=fdAm\Delta f=f_dA_m,調變指數為

β=Δffs=fdAmfs.\beta=\frac{\Delta f}{f_s}=\frac{f_dA_m}{f_s}.

在高載波雜訊比的標準 FM 分析下,單音 FM 的解調訊噪比與 β2\beta^2 成正比;相對於解調前載波雜訊比的改善因子為

SNRoutSNRin=32β2.\frac{\mathrm{SNR}_{\mathrm{out}}}{\mathrm{SNR}_{\mathrm{in}}} =\frac{3}{2}\beta^2.

解題方法

🔒

後續完整解題步驟與【答案】

免費註冊,享三天全站完整詳解閱覽。

免費註冊

第 3 題10 分

  1. (10%) An AM modulator operates with the message signal m(t)=4cos⁡(10πt)+6cos⁡(30πt)m(t) = 4 \cos(10\pi t) + 6 \cos(30\pi t). The unmodulated carrier is 200cos⁡(600πt)200 \cos(600\pi t). The system operates with a modulation index of 0.8.
    (a) Write the equation for mn(t)m_n(t), the normalized signal with a minimum value of -1. (3%)
    (b) Determine <mn(t)><m_n(t)>, the power in mn(t)m_n(t). (4%)
    (c) Determine the efficiency of the modulator. (3%)

登入後即可作答並保存紀錄。

這一題的完整詳解

本題考驗幅度調變 (AM) 系統的相關計算,包含訊號正規化、功率計算及效率分析。

核心觀念:

  • AM 調變的數學模型。
  • 訊號正規化。
  • 訊號功率的計算。
  • AM 調變器的效率。

給定訊息訊號 m(t)=4cos⁡(10πt)+6cos⁡(30πt)m(t) = 4 \cos(10\pi t) + 6 \cos(30\pi t)。
載波訊號為 c(t)=200cos⁡(600πt)c(t) = 200 \cos(600\pi t)。
調變指數 μ=0.8\mu = 0.8。

AM 調變器的輸出訊號一般形式為:
s(t)=Ac[1+μmn(t)]cos⁡(2πfct)s(t) = A_c [1 + \mu m_n(t)] \cos(2\pi f_c t)
其中 AcA_c 是載波幅度,fcf_c 是載波頻率,μ\mu 是調變指數,mn(t)m_n(t) 是正規化的訊息訊號。
由 c(t)=200cos⁡(600πt)c(t) = 200 \cos(600\pi t) 可知,載波幅度 Ac=200A_c = 200,載波角頻率 ωc=600π\omega_c = 600\pi,所以載波頻率 fc=300f_c = 300 Hz。
由 m(t)m(t) 的頻率,10π10\pi 和 30π30\pi 應為角頻率。
m(t)=4cos⁡(ω1t)+6cos⁡(ω2t)m(t) = 4 \cos(\omega_1 t) + 6 \cos(\omega_2 t),其中 ω1=10π\omega_1 = 10\pi rad/s,ω2=30π\omega_2 = 30\pi rad/s。
對應的頻率為 f1=ω1/(2π)=5f_1 = \omega_1 / (2\pi) = 5 Hz,f2=ω2/(2π)=15f_2 = \omega_2 / (2\pi) = 15 Hz。
載波頻率 fc=300f_c = 300 Hz,遠大於訊息訊號的頻率,符合 AM 調變的要求。

(a) Write the equation for mn(t)m_n(t), the normalized signal with a minimum value of -1.

正規化訊息訊號 mn(t)m_n(t) 的定義是將原始訊息訊號 m(t)m(t) 的最大值和最小值分別映射到 1 和 -1。
mn(t)=m(t)−mavgmax⁡∣m(t)−mavg∣m_n(t) = \frac{m(t) - m_{avg}}{\max|m(t) - m_{avg}|}。
如果訊號的平均值 mavg=0m_{avg}=0,則 mn(t)=m(t)max⁡∣m(t)∣m_n(t) = \frac{m(t)}{\max|m(t)|}。
首先,我們需要找到 m(t)m(t) 的最大絕對值。
m(t)=4cos⁡(10πt)+6cos⁡(30πt)m(t) = 4 \cos(10\pi t) + 6 \cos(30\pi t)。
由於 10πt10\pi t 和 30πt30\pi t 的頻率不同,它們的最大值不會同時發生。
因此,最大絕對值通常是各項最大絕對值之和(對於非同步訊號)。
∣m(t)∣max=∣4∣+∣6∣=4+6=10|m(t)|_{max} = |4| + |6| = 4 + 6 = 10。
所以,正規化訊號為:
mn(t)=m(t)10=4cos⁡(10πt)+6cos⁡(30πt)10m_n(t) = \frac{m(t)}{10} = \frac{4 \cos(10\pi t) + 6 \cos(30\pi t)}{10}
mn(t)=0.4cos⁡(10πt)+0.6cos⁡(30πt)m_n(t) = 0.4 \cos(10\pi t) + 0.6 \cos(30\pi t)。
這個正規化訊號的最大值為 0.4+0.6=10.4 + 0.6 = 1,最小值為 −(0.4+0.6)=−1-(0.4 + 0.6) = -1。

【答案】
mn(t)=0.4cos⁡(10πt)+0.6cos⁡(30πt)m_n(t) = 0.4 \cos(10\pi t) + 0.6 \cos(30\pi t)。

(b) Determine <mn(t)><m_n(t)>, the power in mn(t)m_n(t).

訊號的功率定義為其均方值的時間平均。
Psignal=<x(t)2>=lim⁡T→∞1T∫−T/2T/2x2(t)dtP_{signal} = <x(t)^2> = \lim_{T \to \infty} \frac{1}{T} \int_{-T/2}^{T/2} x^2(t) dt。
對於週期性訊號或均值為零的隨機訊號,功率計算會簡化。
對於 mn(t)=0.4cos⁡(10πt)+0.6cos⁡(30πt)m_n(t) = 0.4 \cos(10\pi t) + 0.6 \cos(30\pi t),這是兩個不同頻率的餘弦訊號之和。
由於兩個頻率 10π10\pi 和 30π30\pi 的比值是整數 (30π/10π=330\pi / 10\pi = 3),因此 mn(t)m_n(t) 是週期性訊號。

🔒

後續完整解題步驟與【答案】

免費註冊,享三天全站完整詳解閱覽。

免費註冊

第 4 題24 分

  1. (24%) Consider M-ary PSK (phase-shift keying) and M-ary FSK (frequency-shift keying).
    (a) For M = 2, plot the receiver structures and find the bit error probabilities for both. (8%)
    (b) For M = 2m where m≥ 2 is an integer, find the approximated symbol error probabilities versus E/No for both (E, is the energy per symbol). (8%)
    (c) For M = 2m where m≥ 2 is an integer, find the approximated bit error probabilities versus Eb/No for both (Et is the energy per bit). Please compare the approximated bit error probabilities of PSK and FSK if we increase the value of m (8%).

登入後即可作答並保存紀錄。

這一題的完整詳解

核心觀念

本題比較兩種 MM 元調變:

  • MM-PSK:以載波相位代表符號。
  • MM-FSK:以互相正交的 MM 個頻率代表符號。

設:

  • M=2mM=2^m;
  • EE:每一個符號的能量;
  • EbE_b:每一個位元的能量;
  • 因為每個符號含有 m=log⁡2Mm=\log_2 M 個位元,所以
    E=mEbE=mE_b
  • 通道為 AWGN,雙邊雜訊功率頻譜密度為 N0/2N_0/2;
  • FSK 採用相干偵測,且各頻率訊號彼此正交;
  • Q(x)Q(x) 定義為
    Q(x)=12π∫x∞e−u2/2 duQ(x)=\frac{1}{\sqrt{2\pi}}\int_x^\infty e^{-u^2/2}\,du

主要使用兩個公式:

  1. 二元判決錯誤率由判決變數的高斯尾端機率得到。
  2. MM 元調變的高信噪比錯誤率,可用聯集界(union bound)近似:
    Ps≈∑j≠iP(i→j)P_s\approx \sum_{j\ne i}P(i\rightarrow j)

(a) M=2M=2 時的接收器與位元錯誤率

1. 二元 PSK:BPSK

BPSK 以兩個相反相位表示兩個位元:

s0(t)=+2EbTbcos⁡(2πfct)s_0(t)=+\sqrt{\frac{2E_b}{T_b}}\cos(2\pi f_ct)

s1(t)=−2EbTbcos⁡(2πfct)s_1(t)=-\sqrt{\frac{2E_b}{T_b}}\cos(2\pi f_ct)

相干接收器先將接收訊號與同步載波相乘,再通過積分器或匹配濾波器,最後以零為門檻判決。

             ┌──────────────┐
r(t) ───────►│ × cos(2πfct) │──► 積分器/匹配濾波器 ──► 判決器
             └──────────────┘                         │
                                                       ├─ y > 0:判定 0
                                                       └─ y < 0:判定 1

取正交基底後,接收判決變數可寫成

y=±Eb+ny=\pm\sqrt{E_b}+n

其中

n∼N(0,N02)n\sim\mathcal{N}\left(0,\frac{N_0}{2}\right)

假設傳送 s0s_0,錯誤事件為 y<0y<0,因此

PbBPSK=Q(EbN0/2)P_b^{\text{BPSK}} =Q\left(\frac{\sqrt{E_b}}{\sqrt{N_0/2}}\right)

得到

PbBPSK=Q(2EbN0)\boxed{ P_b^{\text{BPSK}} =Q\left(\sqrt{\frac{2E_b}{N_0}}\right) }

2. 二元 FSK:相干 BFSK

BFSK 使用兩個互相正交的載波:

s0(t)=2EbTbcos⁡(2πf0t)s_0(t)=\sqrt{\frac{2E_b}{T_b}}\cos(2\pi f_0t)

s1(t)=2EbTbcos⁡(2πf1t)s_1(t)=\sqrt{\frac{2E_b}{T_b}}\cos(2\pi f_1t)

其中

∫0Tbs0(t)s1(t) dt=0\int_0^{T_b}s_0(t)s_1(t)\,dt=0

接收器使用兩個相關器,分別計算接收訊號在兩個頻率基底上的投影,再比較兩個輸出值。

                              ┌─► 相關器 f0 ─► y0 ─┐
r(t) ─────────────────────────┤                    ├─► 比較器
                              └─► 相關器 f1 ─► y1 ─┘

                    y0 > y1:判定 0
                    y1 > y0:判定 1

假設傳送 s0s_0,兩個相關器輸出為

y0=Eb+n0y_0=\sqrt{E_b}+n_0

y1=n1y_1=n_1

其中 n0,n1n_0,n_1 互相獨立,且

n0,n1∼N(0,N02)n_0,n_1\sim\mathcal{N}\left(0,\frac{N_0}{2}\right)

判決錯誤條件為 y1>y0y_1>y_0,亦即

y1−y0>0y_1-y_0>0

令

z=y1−y0=−Eb+(n1−n0)z=y_1-y_0=-\sqrt{E_b}+(n_1-n_0)

由於

n1−n0∼N(0,N0)n_1-n_0\sim\mathcal{N}(0,N_0)

所以

PbBFSK=Q(EbN0)P_b^{\text{BFSK}} =Q\left(\frac{\sqrt{E_b}}{\sqrt{N_0}}\right)

因此

Pbcoherent BFSK=Q(EbN0)\boxed{ P_b^{\text{coherent BFSK}} =Q\left(\sqrt{\frac{E_b}{N_0}}\right) }

比較可知,在相同 Eb/N0E_b/N_0 下:

Q(2EbN0)<Q(EbN0)Q\left(\sqrt{\frac{2E_b}{N_0}}\right) < Q\left(\sqrt{\frac{E_b}{N_0}}\right)

所以相干 BPSK 的位元錯誤率低於相干 BFSK。


(b) M=2mM=2^m、m≥2m\ge 2 時的近似符號錯誤率

1. MM-PSK

MM-PSK 的相鄰星座點相位差為

Δθ=2πM\Delta\theta=\frac{2\pi}{M}

每個符號的判決區域由相鄰兩條角平分線構成,因此高信噪比時,主要錯誤來自兩個相鄰符號。

相鄰兩個 PSK 訊號的歐氏距離為

dmin⁡=2Esin⁡(πM)d_{\min}=2\sqrt{E}\sin\left(\frac{\pi}{M}\right)

AWGN 中兩個訊號的成對錯誤機率為

P(i→j)=Q(dmin⁡2N0)P(i\rightarrow j) =Q\left(\frac{d_{\min}}{\sqrt{2N_0}}\right)

代入 dmin⁡d_{\min}:

P(i→j)=Q(2EN0sin⁡πM)P(i\rightarrow j) =Q\left( \sqrt{\frac{2E}{N_0}} \sin\frac{\pi}{M} \right)

每個符號有兩個主要相鄰錯誤,因此

PsM-PSK≈2Q(2EN0sin⁡πM)\boxed{ P_s^{\text{M-PSK}} \approx 2Q\left( \sqrt{\frac{2E}{N_0}} \sin\frac{\pi}{M} \right) }

令 M=2mM=2^m,可寫為

PsM-PSK≈2Q(2EN0sin⁡π2m)\boxed{ P_s^{\text{M-PSK}} \approx 2Q\left( \sqrt{\frac{2E}{N_0}} \sin\frac{\pi}{2^m} \right) }

此式是高信噪比近似;低信噪比時,非相鄰符號錯誤不可忽略。


2. 相干正交 MM-FSK

相干正交 MM-FSK 使用 MM 個互相正交的訊號,每一個訊號能量皆為 EE。

假設傳送第 ii 個訊號,接收器有 MM 個相關器:

                         ┌─► 相關器 1 ─► y1 ─┐
                         ├─► 相關器 2 ─► y2 ─┤
r(t) ────────────────────┼─►    ⋮       ─► ⋮  ├─► 選最大值
                         └─► 相關器 M ─► yM ─┘

接收端選擇輸出最大者:

🔒

後續完整解題步驟與【答案】

免費註冊,享三天全站完整詳解閱覽。

免費註冊

第 5 題8 分

  1. (8%) Consider QPSK (quadriphase-shift keying) and 16-QAM (quadrature-amplitude modulation). Please compare their symbol error probabilities and null-to-null bandwidths in the case of the same data rate.

登入後即可作答並保存紀錄。

這一題的完整詳解

本題考驗 QPSK 和 16-QAM 的符號錯誤機率和頻寬比較。

核心觀念:

  • QPSK 和 16-QAM 的訊號星座圖。
  • 符號錯誤機率 (Symbol Error Probability, PsP_s) 的計算。
  • 頻寬的計算,特別是 null-to-null bandwidth。
  • 相同資料率下的比較。

1. QPSK (Quadrature Phase-Shift Keying)

  • M=4M=4 個符號。
  • 每個符號代表 log⁡2(4)=2\log_2(4) = 2 位元。
  • 星座圖:四個點位於圓上,相位分別為 π/4,3π/4,5π/4,7π/4\pi/4, 3\pi/4, 5\pi/4, 7\pi/4 (或 0,π/2,π,3π/20, \pi/2, \pi, 3\pi/2 等,取決於定義)。
  • 符號錯誤機率 (近似):Ps(QPSK)≈2Q(EsN0sin⁡(π4))=2Q(Es2N0)P_s(\text{QPSK}) \approx 2 Q\left(\sqrt{\frac{E_s}{N_0}} \sin(\frac{\pi}{4})\right) = 2 Q\left(\sqrt{\frac{E_s}{2N_0}}\right)。
    • 這裡的 EsE_s 是每符號能量。
    • 若以位元錯誤機率 PbP_b 表示,則 Es=2EbE_s = 2 E_b。
    • Pb(QPSK)≈2Q(2Eb2N0)=2Q(EbN0)P_b(\text{QPSK}) \approx 2 Q\left(\sqrt{\frac{2E_b}{2N_0}}\right) = 2 Q\left(\sqrt{\frac{E_b}{N_0}}\right)。
  • 頻寬:
    QPSK 使用兩個獨立的 BPSK 訊號(I 和 Q 兩個通道)來傳送兩位元資訊。
    每個通道的頻寬與 BPSK 相同。
    若資料率為 RbR_b bps,則每位元週期 Tb=1/RbT_b = 1/R_b。
    每個通道的符號率為 Rs=Rb/2R_s = R_b / 2。
    每個通道的符號週期 Ts=1/Rs=2TbT_s = 1/R_s = 2 T_b。
    QPSK 的頻譜主要集中在 RsR_s 附近。
    Null-to-null bandwidth 通常是指主瓣的寬度。
    對於使用 Raised Cosine 濾波器的系統,其 null-to-null bandwidth 與符號率 RsR_s 有關。
    若使用 Ideal Nyquist filter (Rectangular pulse shaping),頻寬為 RsR_s。
    考慮到 QPSK 是兩個 BPSK 訊號,每個 BPSK 的頻寬約為 RbR_b。
    若將兩個 BPSK 訊號併合,其頻寬也大約是 RbR_b。
    Null-to-null bandwidth 通常定義為 2Rs2 R_s。
    Rs=Rb/2R_s = R_b / 2。
    所以,Null-to-null bandwidth ≈2Rs=2(Rb/2)=Rb\approx 2 R_s = 2 (R_b / 2) = R_b。
    或者,更精確地說,由於 I 和 Q 訊號共享頻譜,整個 QPSK 訊號的頻寬大約是 RbR_b。
    一些定義將 null-to-null bandwidth 定義為 2Rs2 R_s。

2. 16-QAM (Quadrature Amplitude Modulation)

  • M=16M=16 個符號。
  • 每個符號代表 log⁡2(16)=4\log_2(16) = 4 位元。
  • 星座圖:16 個點,通常排列成一個 4×44 \times 4 的網格。
  • 符號錯誤機率 (近似):對於 square QAM,
    Ps(16-QAM)≈4Q(EsN012.5)P_s(\text{16-QAM}) \approx 4 Q\left(\sqrt{\frac{E_s}{N_0}} \frac{1}{\sqrt{2.5}}\right) (假設 EsE_s 是平均符號能量)。
    更精確的公式是:Ps≈4Q(d/2σ)P_s \approx 4 Q\left(\frac{d/2}{\sigma}\right),其中 dd 是最小距離,σ2=N0/2\sigma^2 = N_0/2。
    對於 square QAM,若符號能量為 EsE_s,則 d=2Es/(5×M/2)d = 2 \sqrt{E_s / (5 \times M / 2)},對於 16-QAM, M=16M=16, EsE_s
    Ps≈4Q(d/2N0/2)=4Q(d2N0)P_s \approx 4 Q\left(\frac{d/2}{\sqrt{N_0/2}}\right) = 4 Q\left(\frac{d}{\sqrt{2N_0}}\right)。
    對於 square QAM,符號能量 Es=M3d2E_s = \frac{M}{3} d^2 (當 M=2mM=2^m, mm 為偶數)。
    Es=163d2  ⟹  d2=3Es16  ⟹  d=3Es16=3Es4E_s = \frac{16}{3} d^2 \implies d^2 = \frac{3 E_s}{16} \implies d = \sqrt{\frac{3 E_s}{16}} = \frac{\sqrt{3 E_s}}{4}。
🔒

後續完整解題步驟與【答案】

免費註冊,享三天全站完整詳解閱覽。

免費註冊

其他考古題

108 年中央大學的其他科目

中央大學《通訊系統》其他年度

其他學校的通訊系考古題