deltacodering

Deltacodering is a data encoding technique in which successive samples are represented by the difference between the current value and a reference value, usually the previous sample. By storing deltas rather than absolute values, the encoding can exploit temporal correlation in signals that change slowly, often reducing the average number of bits needed per sample. The decoder reconstructs the original sequence by cumulatively adding the deltas to the running reference value, assuming synchronization is preserved.

Delta coding is a form of differential encoding and is closely related to differential pulse-code modulation (DPCM). In typical implementations, the delta is computed as current minus previous, and the resulting values may be stored or transmitted as is or quantized. To further compress the data, deltas can be encoded with fixed-length or variable-length codes and combined with entropy coding such as Huffman or arithmetic coding.

Applications include audio and voice transmission, image and video preprocessing where temporal redundancy exists, and time-series data compression in sensing systems. It is also used in streaming protocols and low-bandwidth communications where simple, low-cost encoders are favored.

Advantages of deltacoding include reduced bitrate for smooth signals, simple encoder and decoder structure, and robust synchronization in some scenarios. Its main limitations are susceptibility to error propagation, where a single error shifts all subsequent reconstructed values, and reduced effectiveness on highly variable data. Quantization and noise can accumulate through the delta domain, requiring occasional resynchronization or error-correction.