摘要 |
A method and apparatus for selecting one of a plurality of nominally identical data streams carried on separate data paths is described. A characteristic of the data streams is described which varies randomly or pseudorandomly with time and the selection one of the data paths is based on an estimated probability of an error in a decision to select one of the data paths in accordance with a hypothesis testing algorithm. The data streams may be at constant bit rate. The characteristic of the data streams which varies randomly or pseudorandomly with time can be the number of data cells carried by a data path in a unit time. This is convenient for a transmission protocol such as ATM. <IMAGE> |