说明:接上一节循环自相关函数和谱相关密度(一)——公式推导
7 BPSK信号谱相关密度函数
7.1 实信号模型
BPSK实信号表达式可以写为
r(t)=y(t)+n(t)r(t) = y(t) + n(t)r(t)=y(t)+n(t)
=s(t)p(t)+n(t)= s(t)p(t) + n(t)=s(t)p(t)+n(t)
=∑n=−∞∞a(nT)q(t−nT−t0)cos(2πf0t+θ)+ n(t)= \sum\limits_{n = - \infty }^\infty {a(nT)q(t - nT - {t_0})} \cos (2\pi {f_0}t + \theta ){\text{ + }}n(t)=n=−∞∑∞a(nT)q(t−nT−t0)cos(2πf0t+θ) + n(t)(22)
其中,t0{t_0}t0为起始时间,TTT为符号速率,a(n)a(n)a(n)为基带符号序列,f0{f_0}f0为载波频率,θ\thetaθ为初始相位,n(t)n(t)n(t)为高斯白噪声,q(t)q(t)q(t)为矩形脉冲,其表达式和傅里叶变换为
q(t)={1,∣t∣≤T/20,∣t∣>T/2q(t)=\left\{\begin{array}{ll}1, & |t| \leq T / 2 \\ 0, & |t|>T / 2\end{array}\right.q(t)={1,0,∣t∣≤T/2∣t∣>T/2 (23)
Q(f)=TSa(πfT)Q(f) = T\operatorname{Sa} (\pi fT)Q(f)=TSa(πfT) (24)
且
s(t)=∑n=−∞∞a(nT)q(t−nT−t0)s(t) = \sum\limits_{n = - \infty }^\infty {a(nT)q(t - nT - {t_0})}s(t)=n=−∞∑∞a(nT)q(t−nT−t0)
=q(t−t0)⊗∑na(t)δ(t−nT)= q(t - {t_0}) \otimes \sum\limits_n {a(t)\delta (t - nT)}=q(t−t0)⊗n∑a(t)δ(t−nT)
=q(t−t0)⊗a^(t)= q(t - {t_0}) \otimes \hat a(t)=q(t−t0)⊗a^(t) (25)
p(t)=cos(2πf0t+θ)p(t) = \cos (2\pi {f_0}t + \theta )p(t)=cos(2πf0t+θ) (26)
由(21)知,基带脉冲序列a(nT)a(nT)a(nT)的谱相关密度函数为
S~aα(f)=1T∑n=−∞∞∑m=−∞∞Saα+ m/T(f−m2T−nT)\tilde S_a^\alpha (f) = \frac{1}{T}\sum\limits_{n = - \infty }^\infty {\sum\limits_{m = - \infty }^\infty {S_a^{\alpha {\text{ + }}m/T}(f - \frac{m}{{2T}} - \frac{n}{T})} }S~aα(f)=T1n=−∞∑∞m=−∞∑∞Saα + m/T(f−2Tm−Tn) (27)
由(19)可知,a(t)a(t)a(t)以周期TTT理想抽样后的谱相关密度函数为
Sa^α(f)=1T2∑n=−∞∞∑m=−∞∞Sa^α+ m/T(f−m2T−nT)S_{\hat a}^\alpha (f) = \frac{1}{{{T^2}}}\sum\limits_{n = - \infty }^\infty {\sum\limits_{m = - \infty }^\infty {S_{\hat a}^{\alpha {\text{ + }}m/T}(f - \frac{m}{{2T}} - \frac{n}{T})} }Sa^α(f)=T21n=−∞∑∞m=−∞∑∞Sa^α + m/T(f−2Tm−Tn) (28)
根据傅里叶变换的时移性质,q(t−t0)q(t - {t_0})q(t−t0)的傅里叶变换为Q(f)e−j2πft0Q(f){e^{ - j2\pi f{t_0}}}Q(f)e−j2πft0,则(5)由可得s(t)s(t)s(t)的谱相关密度函数为
Ssα(f)=1TQ(f+α/2)Q∗(f−α/2)e−j2παt0S~aα(f)S_s^\alpha (f) = \frac{1}{T}Q(f + \alpha /2){Q^*}(f - \alpha /2){e^{ - j2\pi \alpha {t_0}}}\tilde S_a^\alpha (f)Ssα(f)=T1Q(f+α/2)Q∗(f−α/2)e−j2παt0S~aα(f) (29)
考虑p(t)p(t)p(t)的二次变换
vτ(t)=p(t+ τ/2)p∗(t−τ/2){v_\tau }(t) = p(t{\text{ + }}\tau /2){p^*}(t - \tau /2)vτ(t)=p(t + τ/2)p∗(t−τ/2)
=14(ej2πf0τ+e−j2πf0τ+ej(4πf0t+2θ)+e−j(4πf0t+2θ))= \frac{1}{4}({e^{j2\pi {f_0}\tau }} + {e^{ - j2\pi {f_0}\tau }} + {e^{j(4\pi {f_0}t + 2\theta )}} + {e^{ - j(4\pi {f_0}t + 2\theta )}})=41(ej2πf0τ+e−j2πf0τ+ej(4πf0t+2θ)+e−j(4πf0t+2θ)) (30)
其Fourier级数系数为
⟨vτ(t)e−j2παt⟩t=14ej2πf0τ⟨e−j2παt⟩t+14e−j2πf0τ⟨e−j2παt⟩t{\left\langle {{v_\tau }(t){e^{ - j2\pi \alpha t}}} \right\rangle _t} = \frac{1}{4}{e^{j2\pi {f_0}\tau }}{\left\langle {{e^{ - j2\pi \alpha t}}} \right\rangle _t} + \frac{1}{4}{e^{ - j2\pi {f_0}\tau }}{\left\langle {{e^{ - j2\pi \alpha t}}} \right\rangle _t}⟨vτ(t)e−j2παt⟩t=41ej2πf0τ⟨e−j2παt⟩t+41e−j2πf0τ⟨e−j2παt⟩t
+14e−j2θ⟨e−j2π(α+2f0)t⟩t+14ej2θ⟨e−j2π(α−2f0)t⟩t+ \frac{1}{4}{e^{ - j2\theta }}{\left\langle {{e^{ - j2\pi (\alpha + 2{f_0})t}}} \right\rangle _t} + \frac{1}{4}{e^{j2\theta }}{\left\langle {{e^{ - j2\pi (\alpha - 2{f_0})t}}} \right\rangle _t}+41e−j2θ⟨e−j2π(α+2f0)t⟩t+41ej2θ⟨e−j2π(α−2f0)t⟩t(31)
则p(t)p(t)p(t)的循环自相关函数和谱相关密度函数为
Rpα(τ)={14e±j2θα=±2f012cos(2πf0τ)α=00otherwise R_{p}^{\alpha}(\tau)=\left\{\begin{array}{cc}\frac{1}{4} e^{\pm j 2 \theta} & \alpha=\pm 2 f_{0} \\ \frac{1}{2} \cos \left(2 \pi f_{0} \tau\right) & \alpha=0 \\ 0 & \text { otherwise }\end{array}\right.Rpα(τ)=⎩⎨⎧41e±j2θ21cos(2πf0τ)0α=±2f0α=0 otherwise (32)
Spα(f)={14e±j2θδ(f)α=±2f014[δ(f+f0)+δ(f−f0)]α=00otherwise S_{p}^{\alpha}(f)=\left\{\begin{array}{cc}\frac{1}{4} e^{\pm j 2 \theta} \delta(f) & \alpha=\pm 2 f_{0} \\ \frac{1}{4}\left[\delta\left(f+f_{0}\right)+\delta\left(f-f_{0}\right)\right] & \alpha=0 \\ 0 & \text { otherwise }\end{array}\right.Spα(f)=⎩⎨⎧41e±j2θδ(f)41[δ(f+f0)+δ(f−f0)]0α=±2f0α=0 otherwise (33)
由(12)、(13)得y(t)y(t)y(t)的循环自相关函数为
Ryα(τ)=∑βRpβ(τ)Rsα−β(τ)R_y^\alpha (\tau ) = \sum\limits_\beta {R_p^\beta (\tau )R_s^{\alpha - \beta }(\tau )}Ryα(τ)=β∑Rpβ(τ)Rsα−β(τ)
=14ej2θRsα−2f0(τ)+14e−j2θRsα+2f0(τ)+12cos(2πf0τ)Rsα(τ)= \frac{1}{4}{e^{j2\theta }}R_s^{\alpha - 2{f_0}}(\tau ) + \frac{1}{4}{e^{ - j2\theta }}R_s^{\alpha + 2{f_0}}(\tau ) + \frac{1}{2}\cos (2\pi {f_0}\tau )R_s^\alpha (\tau )=41ej2θRsα−2f0(τ)+41e−j2θRsα+2f0(τ)+21cos(2πf0τ)Rsα(τ) (34)
Syα(f)=∑βSpβ(f)⊗Ssα−β(f)S_y^\alpha (f) = \sum\limits_\beta {S_p^\beta (f) \otimes S_s^{\alpha - \beta }(f)}Syα(f)=β∑Spβ(f)⊗Ssα−β(f)
=14[Ssα(f+f0)+Ssα(f−f0)+ej2θSsα−2f0(f)+e−j2θSsα+2f0(f)]= \frac{1}{4}\left[ {S_s^\alpha (f + {f_0}) + S_s^\alpha (f - {f_0}) + {e^{j2\theta }}S_s^{\alpha - 2{f_0}}(f) + {e^{ - j2\theta }}S_s^{\alpha + 2{f_0}}(f)} \right]=41[Ssα(f+f0)+Ssα(f−f0)+ej2θSsα−2f0(f)+e−j2θSsα+2f0(f)] (35)
将(29)代入(35),得y(t)y(t)y(t)的谱相关密度函数为
Syα(f)=14T{[Q(f+f0+α/2)Q∗(f+f0−α/2)S~aα(f+f0)S_y^\alpha (f) = \frac{1}{{4T}}\{ [Q(f + {f_0} + \alpha /2){Q^*}(f + {f_0} - \alpha /2)\tilde S_a^\alpha (f + {f_0})Syα(f)=4T1{[Q(f+f0+α/2)Q∗(f+f0−α/2)S~aα(f+f0)
Q(f−f0+α/2)Q∗(f−f0−α/2)S~aα(f−f0)]e−j2παt0Q(f - {f_0} + \alpha /2){Q^*}(f - {f_0} - \alpha /2)\tilde S_a^\alpha (f - {f_0})]{e^{ - j2\pi \alpha {t_0}}}Q(f−f0+α/2)Q∗(f−f0−α/2)S~aα(f−f0)]e−j2παt0
Q(f+f0+α/2)Q∗(f−f0−α/2)S~aα+2f0(f)e−j[2π(α+2f0)t0+2θ]Q(f + {f_0} + \alpha /2){Q^*}(f - {f_0} - \alpha /2)\tilde S_a^{\alpha + 2{f_0}}(f){e^{ - j[2\pi (\alpha + 2{f_0}){t_0} + 2\theta ]}}Q(f+f0+α/2)Q∗(f−f0−α/2)S~aα+2f0(f)e−j[2π(α+2f0)t0+2θ]
Q(f−f0+α/2)Q∗(f+f0−α/2)S~aα−2f0(f)e−j[2π(α−2f0)t0−2θ]}Q(f - {f_0} + \alpha /2){Q^*}(f + {f_0} - \alpha /2)\tilde S_a^{\alpha - 2{f_0}}(f){e^{ - j[2\pi (\alpha - 2{f_0}){t_0} - 2\theta ]}}\}Q(f−f0+α/2)Q∗(f+f0−α/2)S~aα−2f0(f)e−j[2π(α−2f0)t0−2θ]} (36)
对于01先验等概的基带符号序列a(n)a(n)a(n),其循环自相关函数为
R~aα(kT)=limN→∞12N+1∑n=−NNa(nT+kT)a(nT)e−j2πα(n+k/2)T\tilde R_a^\alpha (kT) = \mathop {\lim }\limits_{N \to \infty } \frac{1}{{2N + 1}}\sum\limits_{n = - N}^N {a(nT + kT)a(nT)} {e^{ - j2\pi \alpha (n + k/2)T}}R~aα(kT)=N→∞lim2N+11n=−N∑Na(nT+kT)a(nT)e−j2πα(n+k/2)T (37)
当且仅当k=0k = 0k=0且α=m/T\alpha = m/Tα=m/T时,R~aα(kT)=R~a(0)\tilde R_a^\alpha (kT) = {\tilde R_a}(0)R~aα(kT)=R~a(0),则其谱相关密度函数为
S~aα(f)={R~a(0)=1,α=m/T0,α≠m/T\tilde{S}_{a}^{\alpha}(f)=\left\{\begin{aligned} \tilde{R}_{a}(0)=1, & \alpha=m / T \\ 0, & \alpha \neq m / T \end{aligned}\right.S~aα(f)={R~a(0)=1,0,α=m/Tα=m/T(38)
对于高斯白噪声n(t)n(t)n(t),当且仅当α=0\alpha = 0α=0时,其谱相关密度函数不为零,则BPSK实信号的谱相关密度函数为
Srα(f)={Syα(f)+Snα(f),α=0Syα(f),α≠0S_{r}^{\alpha}(f)=\left\{\begin{array}{cc}S_{y}^{\alpha}(f)+S_{n}^{\alpha}(f), & \alpha=0 \\ S_{y}^{\alpha}(f), & \alpha \neq 0\end{array}\right.Srα(f)={Syα(f)+Snα(f),Syα(f),α=0α=0(39)
由此可见,噪声n(t)n(t)n(t)只影响循环频率为零时的截面。
7.2 复信号模型
BPSK复信号表达式可以写为
r(t)=y(t)+n(t)r(t) = y(t) + n(t)r(t)=y(t)+n(t)
= s(t)p(t)+n(t){\text{ = }}s(t)p(t) + n(t) = s(t)p(t)+n(t)
=∑n=−∞∞a(nT)q(t−nT−t0)ej(2πf0t+θ)= \sum\limits_{n = - \infty }^\infty {a(nT)q(t - nT - {t_0})} {e^{j(2\pi {f_0}t + \theta )}}=n=−∞∑∞a(nT)q(t−nT−t0)ej(2πf0t+θ) (40)
同理,t0{t_0}t0为起始时间,TTT为符号速率,a(n)a(n)a(n)为基带符号序列,f0{f_0}f0为载波频率,θ\thetaθ为初始相位,n(t)n(t)n(t)为高斯白噪声,q(t)q(t)q(t)为矩形脉冲。令
p(t)=ej(2πf0t+θ)p(t) = {e^{j(2\pi {f_0}t + \theta )}}p(t)=ej(2πf0t+θ) (41)
同实数信号模型对比,只有p(t)p(t)p(t)发生了改变,其二次变换的其傅里叶级数系数为
⟨vτ(t)e−j2παt⟩t=⟨p(t+ τ/2)p∗(t−τ/2)e−j2παt⟩t{\left\langle {{v_\tau }(t){e^{ - j2\pi \alpha t}}} \right\rangle _t} = {\left\langle {p(t{\text{ + }}\tau /2){p^*}(t - \tau /2){e^{ - j2\pi \alpha t}}} \right\rangle _t}⟨vτ(t)e−j2παt⟩t=⟨p(t + τ/2)p∗(t−τ/2)e−j2παt⟩t
=ej2πf0τ⟨e−j2παt⟩t= {e^{j2\pi {f_0}\tau }}{\left\langle {{e^{ - j2\pi \alpha t}}} \right\rangle _t}=ej2πf0τ⟨e−j2παt⟩t (42)
则p(t)p(t)p(t)的循环自相关函数和谱相关密度函数为
Rpα(τ)={ej2πf0τα=00α≠0R_{p}^{\alpha}(\tau)=\left\{\begin{array}{cc}e^{j 2 \pi f_{0} \tau} & \alpha=0 \\ 0 & \alpha \neq 0\end{array}\right.Rpα(τ)={ej2πf0τ0α=0α=0(43)
Spα(f)={δ(f−f0)α=00α≠0S_{p}^{\alpha}(f)=\left\{\begin{array}{cc}\delta\left(f-f_{0}\right) & \alpha=0 \\ 0 & \alpha \neq 0\end{array}\right.Spα(f)={δ(f−f0)0α=0α=0(44)
由(12)、(13)得y(t)y(t)y(t)的循环自相关函数为
Ryα(τ)=∑βRpβ(τ)Rsα−β(τ)=ej2πf0τRsα(τ)R_y^\alpha (\tau ) = \sum\limits_\beta {R_p^\beta (\tau )R_s^{\alpha - \beta }(\tau )} = {e^{j2\pi {f_0}\tau }}R_s^\alpha (\tau )Ryα(τ)=β∑Rpβ(τ)Rsα−β(τ)=ej2πf0τRsα(τ) (45)
Syα(f)=∑βSpβ(f)⊗Ssα−β(f)S_y^\alpha (f) = \sum\limits_\beta {S_p^\beta (f) \otimes S_s^{\alpha - \beta }(f)}Syα(f)=β∑Spβ(f)⊗Ssα−β(f)
=δ(f−f0)⊗Ssα(f)= \delta (f - {f_0}) \otimes S_s^\alpha (f)=δ(f−f0)⊗Ssα(f)
=Ssα(f−f0)= S_s^\alpha (f - {f_0})=Ssα(f−f0) (46)
将(29)代入(46),得y(t)y(t)y(t)的谱相关密度函数为
Syα(f)=1T[Q(f−f0+α/2)Q∗(f−f0−α/2)e−j2παt0S~aα(f−f0)]S_y^\alpha (f) = \frac{1}{T}[Q(f - {f_0} + \alpha /2){Q^*}(f - {f_0} - \alpha /2){e^{ - j2\pi \alpha {t_0}}}\tilde S_a^\alpha (f - {f_0})]Syα(f)=T1[Q(f−f0+α/2)Q∗(f−f0−α/2)e−j2παt0S~aα(f−f0)] (47)
同(39),可得复BPSK信号的谱相关密度函数为
Srα(f)={Syα(f)+Snα(f),α=0Syα(f),α≠0S_{r}^{\alpha}(f)=\left\{\begin{array}{cc}S_{y}^{\alpha}(f)+S_{n}^{\alpha}(f), & \alpha=0 \\ S_{y}^{\alpha}(f), & \alpha \neq 0\end{array}\right.Srα(f)={Syα(f)+Snα(f),Syα(f),α=0α=0(48)
7.3 谱分析
7.3.1 主峰个数
对于实BPSK信号,由(36)、(38)可知,其谱相关密度函数在f=0f = 0f=0且α=±2f0\alpha = \pm 2{f_0}α=±2f0处各有一个主峰;在α=0\alpha = 0α=0且f=±f0f = \pm {f_0}f=±f0处各有一个主峰,即实BPSK信号共有4个主峰。
对于复BPSK信号,由(47)、(48)可知,其谱相关密度函数只有在f=f0f = {f_0}f=f0且α=0\alpha = 0α=0处有一个谱峰。
7.3.2 切面特征
在式(36)中,令f=0f = 0f=0,α=±2f0+m/T\alpha = \pm 2{f_0} + m/Tα=±2f0+m/T得
Syα(f)={14T∣Q(−f0+α/2)∣2e−j[2πnt0/T−2θ]α=2f0+m/T14T∣Q(f0+α/2)∣2e−j[2πnt0/T+2θ]α=−2f0+m/TS_{y}^{\alpha}(f)=\left\{\begin{array}{ll}\frac{1}{4 T}\left|Q\left(-f_{0}+\alpha / 2\right)\right|^{2} e^{-j\left[2 \pi n t_{0} / T-2 \theta\right]} & \alpha=2 f_{0}+m / T \\ \frac{1}{4 T}\left|Q\left(f_{0}+\alpha / 2\right)\right|^{2} e^{-j\left[2 \pi n t_{0} / T+2 \theta\right]} & \alpha=-2 f_{0}+m / T\end{array}\right.Syα(f)={4T1∣Q(−f0+α/2)∣2e−j[2πnt0/T−2θ]4T1∣Q(f0+α/2)∣2e−j[2πnt0/T+2θ]α=2f0+m/Tα=−2f0+m/T(49)
特别地,当m=0m = 0m=0时,有
Syα(f)={14T∣Q(0)∣2ej2θα=2f014T∣Q(0)∣2e−j2θα=−2f0S_{y}^{\alpha}(f)=\left\{\begin{array}{ll}\frac{1}{4 T}|Q(0)|^{2} e^{j 2 \theta} & \alpha=2 f_{0} \\ \frac{1}{4 T}|Q(0)|^{2} e^{-j 2 \theta} & \alpha=-2 f_{0}\end{array}\right.Syα(f)={4T1∣Q(0)∣2ej2θ4T1∣Q(0)∣2e−j2θα=2f0α=−2f0(50)
即在f=0f = 0f=0切面,其谱相关密度函数幅度最大值出现在循环频率为α=±2f0\alpha = \pm 2{f_0}α=±2f0处,由此可估计实BPSK信号的载波频率;在其左右偏移符号速率处,出现次峰值,可估计其符号速率,且可根据α=±2f0\alpha = \pm 2{f_0}α=±2f0处对应的谱相关密度函数的相位来估计初相θ\thetaθ。
令f=±f0f = \pm {f_0}f=±f0,α=m/T\alpha = m/Tα=m/T得
Syα(f)=14T{[Q(2f0+α/2)Q∗(2f0−α/2)+∣Q(α/2)∣2]e−j2παt0S_y^\alpha (f) = \frac{1}{{4T}}\{ [Q(2{f_0} + \alpha /2){Q^*}(2{f_0} - \alpha /2) + |Q(\alpha /2){|^2}]{e^{ - j2\pi \alpha {t_0}}}Syα(f)=4T1{[Q(2f0+α/2)Q∗(2f0−α/2)+∣Q(α/2)∣2]e−j2παt0 (51)
即在f=±f0f = \pm {f_0}f=±f0切面,其谱相关密度函数幅度在循环频率为α=m/T\alpha = m/Tα=m/T即符号速率整数倍处出现峰值,在α=0\alpha = 0α=0处的峰值最大,由此可估计实BPSK信号的符号速率,此外还可根据符号速率处对应的谱相关密度函数的相位来估计时延t0{t_0}t0;其中,需要注意的是,当频率分辨率远远小于循环频率分辨率,即Δf≫Δα\Delta f \gg \Delta \alphaΔf≫Δα时,符号速率处对应的峰值才比较明显。
对于复BPSK信号,在式(47)中,令α=0\alpha = 0α=0,得
Syα(f)=1T∣Q(f−f0)|2S_y^\alpha (f) = \frac{1}{T}|Q(f - {f_0}){{\text{|}}^2}Syα(f)=T1∣Q(f−f0)|2 (52)
即在α=0\alpha = 0α=0切面,其谱相关密度函数幅度只在f=f0f = {f_0}f=f0出现峰值,由此可估计复BPSK信号的载波频率,但此时噪声n(t)n(t)n(t)的谱相关密度函数不为零,因此利用该切面进行载频估计受噪声影响较大。
令f=f0f = {f_0}f=f0,得
Syα(f)=1T∣Q(α/2)∣2e−j2παt0S~aα(0)S_y^\alpha (f) = \frac{1}{T}|Q(\alpha /2){|^2}{e^{ - j2\pi \alpha {t_0}}}\tilde S_a^\alpha (0)Syα(f)=T1∣Q(α/2)∣2e−j2παt0S~aα(0) (53)
即在f=f0f = {f_0}f=f0切面,其谱相关密度函数幅度在循环频率为α=m/T\alpha = m/Tα=m/T即符号速率整数倍处出现峰值,在α=0\alpha = 0α=0处的峰值最大,由此可估计实BPSK信号的符号速率,此外还可根据符号速率处对应的谱相关密度函数的相位来估计时延t0{t_0}t0。
7.4 成形滤波器对谱相关密度函数的影响
无论是BPSK还是QPSK调制信号,对于矩形成形,其频谱为Sa函数,当∣f∣>1/T\left| f \right| > 1/T∣f∣>1/T时,存在衰减较慢的旁瓣,因此在循环频率为α=m/T\alpha = m/Tα=m/T或α=m/T±2f0\alpha = m/T \pm 2{f_0}α=m/T±2f0处其谱相关密度函数仍然不为零,即在主峰周围会有很多小峰。对于(根)升余弦成形,当∣f∣>1/T\left| f \right| > 1/T∣f∣>1/T时,其频谱较快衰减为零,因此其谱相关密度函数只在循环频率为α=1/T\alpha = 1/Tα=1/T或α=1/T±2f0\alpha = 1/T \pm 2{f_0}α=1/T±2f0处有值。