Why my frequency domain graph looks so weird?

This is the result of my frequency modulated signal in frequency domain, may I ask whether how can I avoid the big gap between each cycle

below is my coding

%FM generation
clc;
clear all;
close all;
fc=input('Enter the carrier signal freq in hz,fc=');
fm=input('Enter the modulating signal freq in hz,fm =');
m=input('Modulation index,m= ');
t=(1/fc)*20;
t=0:0.0001:t;
c=cos(2*pi*fc*t);%carrier signal
M=sin(2*pi*fm*t);% modulating signal
subplot(3,1,1);plot(t,c);
ylabel('amplitude');xlabel('time index');title('Carrier signal');
subplot(3,1,2);plot(t,M);
ylabel('amplitude');xlabel('time index');title('Modulating signal');
y=cos(2*pi*fc*t-(m.*cos(2*pi*fm*t)));
subplot(3,1,3);plot(t,y);
ylabel('amplitude');xlabel('time index');title('Frequency Modulated signal');

fs=1000;% this will define the the resolution.
z=fft(y);
nfft = length(y);
f = (0:1/nfft:1-1/nfft)*fs; % define frequency-domain
figure;         % figure should be written before subplot to open new figure
subplot(2,1,1); % subplot(2,1,4) will give error beacause for a 2x1 vector valid indeces are 1&2, 4 is wrong
plot(f,z);      % t (time-domain) is replace with f (frequency-domain)
ylabel('amplitude');xlabel('frequency domain');title('Frequency Modulated signal');

[Walter Roberson]

What are the inputs you are using in response to the input() requests?

Answer 1

[Star Strider]

You are plotting only the real part of a two-sided Fourier transform.

Try this:

fs=1000;% this will define the the resolution.
fn = fs/2;          % Nyquist Frequency
z=fft(y)/numel(y);
nfft = fix(length(y)/2);
f = (0:1/nfft:1-1/nfft)*fn; % define frequency-domain
figure;         % figure should be written before subplot to open new figure
subplot(2,1,1); % subplot(2,1,4) will give error beacause for a 2x1 vector valid indeces are 1&2, 4 is wrong
plot(f,abs(z(1:nfft)));      % t (time-domain) is replace with f (frequency-domain)
ylabel('amplitude');xlabel('frequency domain');title('Frequency Modulated signal');