1: \begin{abstract}

2: In this paper, we consider weighted nonbinary repeat

3: multiple-accumulate (WNRMA) code ensembles obtained from the serial

4: concatenation of a nonbinary rate-$1/n$ repeat code and the cascade

5: of $L\geq 1$ accumulators, where each encoder is followed by a

6: nonbinary random weighter. The WNRMA codes are assumed to be

7: iteratively decoded using the turbo principle with maximum \emph{a

8: posteriori} constituent decoders. We derive the exact weight

9: enumerator of nonbinary accumulators and subsequently give the

10: weight enumerators for WNRMA code ensembles. We formally prove that

11: the symbol-wise minimum distance of WNRMA code ensembles

12: asymptotically grows linearly with the block length when $L \geq 3$

13: and $n \geq 2$, and $L=2$ and $n \geq 3$, for all powers of primes

14: $q \geq 3$ considered, where $q$ is the field size. Thus, WNRMA code

15: ensembles are \emph{asymptotically good} for these parameters. We

16: also give iterative decoding thresholds, computed by an extrinsic

17: information transfer chart analysis,  on the $q$ary symmetric

18: channel to show the convergence properties. Finally, we consider the

19: binary image of WNRMA code ensembles and compare the asymptotic

20: minimum distance growth rates with those of binary repeat

21: multiple-accumulate code ensembles.

22: \end{abstract}

23: