In this thesis, we propose and evaluate quasi-optimal algorithms for solving the code-constellation optimization problem in M-ary CDMA system. The M-ary CDMA system is a new CDMA architecture. The more spreading codes used in each user, and the higher bandwidth efficiency can achieve with more bits packed in each symbol. We use a code, which we refer to as ¡§mapping code¡¨, to help form a multidimensional spherical code-constellation. The M codewords of the mapping code correspond one-to-one to the M points on the code-constellation. Thus, the code-constellation optimization problem is a combinatorial optimization problem. We present that an exhaustive search (ES) algorithm would have compute and check all possible subset, and then this problem becomes a NP-hard. Based on the exhaustive search algorithm, we propose symmetric points search (SPS) algorithm to reduce computation complexity, but it is not optimal algorithm. In addition, we propose a quasi-optimal algorithm, namely Manhattan distance search (MDS) algorithm. Numerical results and comparisons are provided to illustrate that the computation complexity of the Manhattan distance search algorithm increases linearly with dimension of code-constellation and its performance is better than others.