next up previous
Next: Time-Varying Channels Up: Collected Problems Previous: Maximum Likelihood Sequence Estimation

Continuous Phase Modulation

Equally likely random symbols $I_n \in \{-1,1\}$ are to be transmitted at a rate of one symbol per $T$ seconds over an additive white Gaussian noise channel. The symbols are modulated using partial response ($L=2$) continuous phase modulation with modulation index $h=\frac{1}{2}$ and rectangular pulses.

  1. Sketch the phase of the transmitted signal resulting from the sequence

    \begin{displaymath}
\underline{I} = \{1,-1,1,1 -1,-1,-1,1\}.
\end{displaymath}

  2. Show that for an arbitrary input sequence $\underline{I}$ the phase can be expressed as

    \begin{displaymath}
\phi(t,\underline{I}) = \theta_n + 2 \pi h \sum_{k=0}^{L-1} I_{n-k}
q(t-(n-k)T) \quad \mbox{for $nT \leq t < (n+1)T$}.
\end{displaymath}

    Provide an explicit expression for $\theta_n$.
  3. Which values can $\theta_n$ take on?
  4. Phase transitions between times $nT$ and $(n+1)T$ depend only on $\theta_n$, $I_n$, and $I_{n-1}$. Hence, the evolution of the phase $\phi(t,\underline{I})$ can be described as a path through a suitably chosen trellis. How would you define the states of the trellis?
  5. Draw a trellis diagram showing all possible state transitions between times $nT$ and $(n+1)T$.
  6. To employ the Viterbi algorithm for demodulating CPM signals the metrics (distances) associated with the state transitions are required. Define these distances precisely and explain how they can be obtained.


next up previous
Next: Time-Varying Channels Up: Collected Problems Previous: Maximum Likelihood Sequence Estimation
Dr. Bernd-Peter Paris
2003-12-08