Summary: The traditional Huffman coding, a variable-length coding technique, is widely used for lossless data compression. A generalized Huffman tree coding algorithm is developed to reduce the time and memory for real-time processing. The layer and probability-table order are used to ensure the uniqueness of coding and decoding. A shift sort is used instead of a search with an indexing table used to simplify the sort. A decoding algorithm is also given. Compared with the traditional Huffman coding technique used in other improved algorithms, the current algorithm is significantly simplifier with less processing time and memory and more balanced coding.