1: \begin{abstract} A   time-invariant, linear, distributed  observer is described for  estimating the state

2: of an $m>0$ channel, $n$-dimensional continuous-time linear system

3: of the form $\dot{x} = Ax,\;y_i = C_ix,\;i\in\{1,2,\ldots, m\}$. The

4: state $x$ is   simultaneously estimated by $m$ agents assuming each

5: agent $i$ senses  $y_i$ and receives the state $z_j$ of each of its

6: neighbors' estimators. Neighbor relations are characterized by a

7: constant directed  graph $\mathbb{N}$ whose

8:  vertices correspond to agents and whose arcs depict neighbor relations. For the case when the neighbor graph is strongly connected, the overall distributed observer consists of

9:  $m$ linear estimators, one for each agent; $m-1$ of the estimators are of dimension  $n$ and one estimator is of dimension $n+m-1$.

10:   Using  results from classical decentralized control theory, it is shown that subject to  the assumptions that

11:    (i) none of the $C_i$ are zero,

12: (ii) the   neighbor graph $\mathbb{N}$ is strongly connected, (iii)

13: the system whose state is to be estimated is jointly observable,

14:  and nothing more,

15:     it is possible to freely assign the spectrum of the overall distributed observer.  For the  more general

16:     case when $\mathbb{N}$ has $q>1$ strongly connected components,

17:     it is  explained how to construct a family of  $q$ distributed

18:     observers, one for each component, which can estimate $x$ at a preassigned convergence rate.

19: \end{abstract}

20: