We propose a new concept of network dimensioning, which is based not only on the grade of service but also on profit. In traditional network dimensioning methodology, the number of circuits on links is designed under a cost-minimization concept with grade of service constraints. Recently, telecommunication markets have become very large and competitive; therefore, we believe that a profit viewpoint is now essential. However, it is difficult to calculate profit in almost all the dimensioning methods currently used, because they mainly employ peak-hour traffic data, while profit depends on all the hourly traffic data which contain both peak and off-peak data. In this paper, we propose using all the hourly traffic data in network dimensioning. From these data and telephone charges for each hour, revenues will be estimated. On the other hand, facility costs will be estimated from the number of circuits. Finally, we can estimate profit from the difference between revenues and facility costs. Focusing on both quality and profits in network dimensioning leads to more advanced quality management and quality control in telecommunications networks than with traditional methodology. This paper outlines a dimensioning method based on profit, and describes its properties, some applications of it, and summarizes further studies.

The stage method used in the analysis of traffic or reliability models causes an increase in the number of state. For example, the k2-stage Erlang distribution is used to approximate the general type of distribution which has a coefficient of variation β1/k (ex. the 100-stage Erlang distribution is used for β0.1). This paper proposes a new type of stage method that needs only two stages to approximate a general distribution with arbitrary coefficient of variation β(0β) and presents some useful applications.