How will the signal $\sum_{n=-\infty}^\infty \Delta (t-n)u_{-1}(t-n)$ look like?

  1. I understand that $\sum_{n=-\infty}^\infty \Delta (t-n)$ is a periodic signal with period $1$, where the triangular pulse $\Delta(t)$ is shifted by $1$ units for each value of $n$.

    where, $$\Delta(t) = \begin{cases} 1-2\vert t\rvert,& \lvert t\rvert<\frac{1}{2} \\ 0,&\lvert t\rvert>\frac{1}{2} \end{cases}$$

  2. From this I infer that $\sum_{n=-\infty}^\infty \Delta (t-n)u(t-n)$ would probably be a signal in which the RIGHT HALF of $\Delta (t)$ will be shifted by $1$ unit for every value of $n$. But here I have a doubt that $u(t-n)$ will be added with the previous values of $u(t-n)$ i.e $u(t-(n-1))$ and will have a staircase effect. Thus the signal will no longer be periodic.(Which is a cause of concern as we have been asked to find the Fourier series of the signal, which means the signal should be periodic)

  3. I also don't understand what the subscript $-1$ in $u_{-1}(t-n)$ means.

Please help me visualize this signal(It would be highly helpful if you would draw the graph as well to help me visualize the signal)

  • 1
    $\begingroup$ As long as we don't know what $u_{-1}(t)$ means I'm afraid nobody can help you. If that exercise is from a book, they should define it somewhere. My guess would be that it's a time-inverted step (but I don't know why one would need such a notation), i.e., $u_{-1}(t)=u(-t)$, but that's just a wild guess. $\endgroup$ – Matt L. Feb 22 '17 at 7:20
  • $\begingroup$ @MattL. I too was thinking of that possibility, but when I saw the Fourier coefficients being calculated as $x_n=\int_{0}^1(-t+1)e^{-j2\pi nt}$ in the solution manual(this one is authorised), I ruled out that possibility.(Their definition of $\Delta (t)$ is slightly different from what I have defined above) Sir, what if the subscript -1 had not been present, what would the case have been? Would my interpretation have been correct in that case? $\endgroup$ – Soumee Feb 22 '17 at 7:35

Initially, don't worry about the sum. Just try to figure out how the signal $\Delta(t)u_{-1}(t)$ looks like. As soon as you know that, you just need to shift it by integer values of $t$, and add them all up. If the shifted versions do not overlap, you simply get copies of the basic signal shifted to integer values of $t$. From the solution in the solution manual (as given in a comment), the basic signal is the right half of a triangle, and the period equals $1$. The symbol $u_{-1}(t)$ refers to the unit step function, which is more frequently referred to as $u(t)$ (see also this question and its answer(s)).


Assuming that you're referring to the conventions used for the unit doublets (see Signals and Systems,2e,Ch.2,Sec 2.5.3, A.Opp) then $u_{-1}(t) = u(t)$, where the right hand side is the conventional unit step (heaviside) function.

Therefore given the definition of that triangular pulse as: $$\Delta(t) = \begin{cases} 1-2|t|, \, |t|<\frac{1}{2} \\ 0,\,|t|>\frac{1}{2} \end{cases}$$

we can see that the required signal is a periodic extension, with period 1, of the following base period : $$\Delta_{B}(t) = \begin{cases} 1-2t, ~~~~ 0<t<\frac{1}{2} \\ 0 ~~~~~,~~~~~~\frac{1}{2} < t <1 \end{cases}$$

The resulting sum can be seen to be the following:enter image description here

Note that the pulses extend from $-\infty$ to $\infty$ ...

  • $\begingroup$ Strange notation ... so $u_0(t)=\delta(t)$?? I prefer $\delta'(t)$ or $\delta^{(n)}(t)$ for the first or n'th derivative of $\delta(t)$. By the way, according to the formula for the Fourier coefficients given in a comment under the question, the half-triangle hits zero at $t=1$, not at $t=1/2$, doesn't it? $\endgroup$ – Matt L. Feb 22 '17 at 10:32
  • $\begingroup$ Yes true. Just a notation. And $u_1(t)$ is $\delta'(t)$ and it has the following properties; $u_m(t) = u_1(t) \star u_1(t) \star ... \star u_1(t)$ m-times, for $m$ pos integer and similarly $u_m(t) = u_{-1}(t) \star u_{-1}(t) \star ... \star u_{-1}(t)$ for $m$ neg int. $u_k(t) \star u_m(t) = u_{k+m} (t)$. He seems to make a typo either in the time domain definition or in the Fourier series integrand. I have choosen the time domain definition to be correct. If the latter is correct then as you pointed, the triangle will hit zero at $t=1$ and the periodic wave will be kind of a sawtooth. $\endgroup$ – Fat32 Feb 22 '17 at 12:01
  • $\begingroup$ At this point I don't understand this discussion(those in the comments,though I have understood the answer by Mr.Fat32),but would like to mention that the Fourier coefficients has been calculated considering the following function of $\Delta (t)$ $$\Delta(t) = \begin{cases} 1-\vert t\rvert,& \lvert t\rvert<=1 \\ 0,&\lvert t\rvert>1 \end{cases}$$ which results in $x_n=\int_{0}^1(-t+1)e^{-j2\pi nt}$ $\endgroup$ – Soumee Feb 22 '17 at 12:49
  • $\begingroup$ Sir, would you mind having a look at this question: dsp.stackexchange.com/q/66740/26549 $\endgroup$ – Soumee Apr 21 '20 at 18:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.