TCP performance over current cellular access: A comprehensive analysis

Eneko Atxutegi, Åke Arvidsson, Fidel Liberal, Karl Johan Grinnemo, Anna Brunstrom

Research output: Chapter in Book/ReportChapter in bookpeer-review

9 Citations (Scopus)
202 Downloads (Pure)

Abstract

Mobile Internet usage has increased significantly over the last decade and it is expected to grow to almost 4 billion users by 2020. Even after the great effort dedicated to improving the performance, there still exist unresolved questions and problems regarding the interaction between TCP and mobile broadband technologies such as LTE. This chapter presents a thorough investigation of the behavior of distinct TCP implementation under various network conditions in different LTE deployments including to which extent TCP is capable of adapting to the rapid variability of mobile networks under different network loads, with distinct flow types, during start-up phase and in mobile scenarios at different speeds. Loss-based algorithms tend to completely fill the queue, creating huge standing queues and inducing packet losses both under stillness and mobility circumstances. On the other side delay-based variants are capable of limiting the standing queue size and decreasing the amount of packets that are dropped in the eNodeB, but under some circumstances they are not able to reach the maximum capacity. Similarly, under mobility in which the radio conditions are more challenging for TCP, the loss-based TCP implementations offer better throughput and are able to better utilize available resources than the delay-based variants do. Finally, CUBIC under highly variable circumstances usually enters congestion avoidance phase prematurely, provoking a slower and longer start-up phase due to the use of Hybrid Slow-Start mechanism. Therefore, CUBIC is unable to efficiently utilize radio resources during shorter transmission sessions.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Chapter14
Pages371-400
Number of pages30
DOIs
Publication statusPublished - 2018-May-25

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
VolumeLNCS 10768
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Swedish Standard Keywords

  • Communication Systems (20203)

Keywords

  • Flow size
  • LTE
  • Mobility
  • Slow-start
  • TCP adaptability

Fingerprint

Dive into the research topics of 'TCP performance over current cellular access: A comprehensive analysis'. Together they form a unique fingerprint.

Cite this