Window functions in MATLAB - Time & Space Complexity

Time Complexity: Window functions

O(N * M)

Understanding Time Complexity

We want to understand how the time needed to run a window function changes as the input size grows.

Specifically, how does the number of calculations increase when we apply a window over a signal?

Scenario Under Consideration

Analyze the time complexity of the following code snippet.

N = length(signal);
window = hamming(M); % window length M
output = zeros(1, N-M+1);
for i = 1:(N-M+1)
    segment = signal(i:i+M-1);
    output(i) = sum(segment .* window);
end
    

This code applies a sliding window of size M over the input signal and computes a weighted sum for each position.

Identify Repeating Operations

Identify the loops, recursion, array traversals that repeat.

• Primary operation: The for-loop runs over the signal, and inside it, the sum of M elements is computed.
• How many times: The outer loop runs about N times, and each iteration does M multiplications and additions.

How Execution Grows With Input

As the signal length N grows, the number of times we slide the window grows roughly with N. For each slide, we do M operations.

Input Size (N)	Approx. Operations
10	About 10 * M
100	About 100 * M
1000	About 1000 * M

Pattern observation: The total work grows linearly with the signal length N, multiplied by the fixed window size M.

Final Time Complexity

Time Complexity: O(N * M)

This means the time needed grows proportionally to the signal length times the window size.

Common Mistake

[X] Wrong: "The time grows only with N because the window size M is small and fixed."

[OK] Correct: Even if M is fixed, it still multiplies the work done for each step, so it affects total time linearly.

Interview Connect

Understanding how window functions scale helps you explain performance in signal processing tasks clearly and confidently.

Self-Check

"What if we used a variable window size that grows with N? How would the time complexity change?"

Practice

1. What is the primary purpose of using a window function in MATLAB signal processing?

easy

A. To reduce edge effects and smooth the signal

B. To increase the signal amplitude

C. To convert the signal from time to frequency domain

D. To generate random noise

Solution

1. Step 1: Understand window function role

   Window functions are used to taper the edges of a signal to reduce artifacts when processing.

2. Step 2: Identify the effect on signal

   They smooth the signal edges and reduce spectral leakage, not amplify or transform the signal.

3. Final Answer:

   To reduce edge effects and smooth the signal -> Option A

4. Quick Check:

   Window functions reduce edge effects [OK]

Hint: Window functions smooth edges to reduce artifacts [OK]

Common Mistakes:

• Thinking window functions amplify signals
• Confusing windowing with Fourier transform
• Assuming window functions add noise

2. Which of the following is the correct MATLAB syntax to create a Hamming window of length 64?

easy

A. hamming(64)

B. hamming[64]

C. window(hamming, 64)

D. window('hamming', 64)

Solution

1. Step 1: Recall MATLAB window function syntax

   MATLAB provides direct functions like hamming(N) to create windows of length N.

2. Step 2: Check each option

   hamming(64) uses correct function call syntax: hamming(64). window('hamming', 64) uses string incorrectly, C misses quotes, B uses wrong brackets.

3. Final Answer:

   hamming(64) -> Option A

4. Quick Check:

   Use hamming(N) to create Hamming window [OK]

Hint: Use hamming(N) directly, not window('hamming', N) [OK]

Common Mistakes:

• Using quotes incorrectly in function calls
• Using square brackets instead of parentheses
• Confusing window function with generic window()

3. What is the output size of the following MATLAB code?

win = blackman(100);
size_win = size(win);

medium

A. [1 100]

B. [100 100]

C. [100 1]

D. [1 1]

Solution

1. Step 1: Understand blackman output shape

   blackman(N) returns a column vector of length N, so size is N rows by 1 column.

2. Step 2: Apply to N=100

   Output is 100 rows and 1 column, so size is [100 1].

3. Final Answer:

   [100 1] -> Option C

4. Quick Check:

   blackman(N) returns column vector size [N 1] [OK]

Hint: Window functions return column vectors by default [OK]

Common Mistakes:

• Assuming output is a row vector
• Confusing size with length
• Expecting square matrix output

4. Identify the error in this MATLAB code snippet:

win = window('hann', 50);
plot(win);

medium

A. The 'hann' window is not supported in MATLAB

B. The plot function cannot plot window vectors

C. The window length must be greater than 100

D. The window function does not accept string arguments

Solution

1. Step 1: Check window function usage

   In MATLAB, window functions like hann are called directly, not via window('hann', N).

2. Step 2: Validate other options

   Plot can plot vectors, length 50 is valid, and 'hann' window is supported directly.

3. Final Answer:

   The window function does not accept string arguments -> Option D

4. Quick Check:

   Use hann(N), not window('hann', N) [OK]

Hint: Call hann(N) directly, not window('hann', N) [OK]

Common Mistakes:

• Using window() with string instead of direct function
• Thinking plot can't display vectors
• Assuming window length limits

5. You want to apply a Hamming window to a noisy signal vector sig of length 256 to reduce spectral leakage before FFT. Which MATLAB code correctly applies the window and computes the FFT?

hard

A. win = hamming(length(sig)); fft_sig = fft(sig + win);

B. win = hamming(256); fft_sig = fft(sig .* win);

C. win = hamming(256); fft_sig = fft(sig / win);

D. win = hamming(256); fft_sig = fft(sig - win);

Solution

1. Step 1: Create window matching signal length

   Use hamming(256) or hamming(length(sig)) to create the window vector.

2. Step 2: Apply window by element-wise multiplication

   Multiply signal by window element-wise to taper edges before FFT.

3. Step 3: Compute FFT of windowed signal

   Use fft(sig .* win) to get frequency spectrum of windowed signal.

4. Final Answer:

   win = hamming(256); fft_sig = fft(sig .* win); -> Option B

5. Quick Check:

   Window applied by multiplication before FFT [OK]

Hint: Multiply signal by window before FFT [OK]

Common Mistakes:

• Adding or subtracting window instead of multiplying
• Dividing signal by window
• Using window length different from signal length