Slidingwindow

Sliding window is a general concept used in computer science to process data in a contiguous window that moves along a sequence. The window is defined by left and right bounds and can be of fixed or variable size. As the window advances, elements enter and leave the window, enabling incremental computation without reprocessing the entire sequence. The approach is commonly implemented with two pointers and supporting data structures such as hash maps to track state.

In algorithm design, the sliding window technique is used to solve problems on arrays or strings efficiently. Typical tasks include finding the maximum or minimum sum of a subarray of length k, finding the longest substring that satisfies a property, or maintaining a count of distinct characters within a moving window. Time complexity is often linear, O(n), with space complexity depending on the properties tracked inside the window.

Sliding window protocol is a data link layer or transport layer flow control mechanism used in reliable transmission over a potentially unreliable channel. The sender can transmit multiple frames before receiving acknowledgments, up to a window size. As acknowledgments arrive, the window slides forward, enabling new frames to be sent. Variants include stop-and-wait (window size 1), go-back-N (upon error, retransmit from the failed frame), and selective repeat (only erroneous frames are retransmitted). The window ensures in-order delivery and regulates the pace of transmission.

Practical implementations must handle loss, errors, and timeouts, and may use timers, sequence numbers, and buffering. The concept is referred to as sliding window in both software algorithms and networking contexts, and though related by the idea of moving bounds, they address different problems: in-data processing versus reliable transmission.