The speed of the data transfer plays an essential role for user experience. The modern Internet adopts the Internet Protocol Stack model, where the transport layer provides important services that severely influence the data transfer rate. In order to achieve new performance gains, network protocols are constantly being improved.

One of the most important new developments at the moment is the new generation transport protocol QUIC. It was originally designed by Google and developed to overcome some of the well-known problems and performance issues in Internet data transfer. It solves, for example, the head of line blocking issue, reduces connection establishment latency, and supports connection migration. QUIC is not only limited to transport-specific services; it also includes data encryption. In contrast to other transport protocols, QUIC is deployed in user space and can be easily extended to prevent version ossification.

Although QUIC includes many advanced options, it suffers from performance degradation compared to TCP. Increased protocol complexity makes it hard to ex- plore non-trivial interactions between new protocol features. One of the open prob- lems in QUIC is how to adapt the acknowledgment (ACK) processing of the novel transport protocol to a wide variety of network configurations. ACK thinning, imple- mented in middleboxes, is widespread for TCP. However, it is not possible for QUIC due to the encrypted header and payload.

In this work, we propose an ACK optimization algorithm for QUIC that can dy- namically adapt the acknowledgment policy to changing network environments and successfully handle this problem. Moreover, we examine the behavior and impact of advanced protocol features on QUIC performance, including flow control, different congestion control mechanisms, and the influence of control frames on an efficient ACK processing strategy.