UČINKOVITOST PROTOKOLA TCP V RAZLIČNIH OMREŽNIH OKOLJIH

BEŠLAGIĆ, SANJIN (author), Hercog, Drago (mentor)

Tema naloge je simulacija protokola TCP v različnih omrežnih okoljih. Specifikacija protokola TCP (transmission control protocol) v osnovnem načinu delovanja brez izgub je podana v formalnem programskem jeziku SDL (specification and description language). Potem je izveden prehod v programsko kodo v programskem jeziku C. Koda v C oz. simulacijski program skupaj s simulatorjem, ki vsebuje različne simulacijske funkcije, med njimi tudi naključne, omogoča izvajanje eksperimentov in uvid v učinkovitost TCP protokola, ko gre za različna izgubna omrežna okolja. Naključne funkcije določajo: a) verjetnost, da se bo kakšen segment pri prenosu izgubil zaradi zamašitev pri naključno generiranem bremenskem prometu v vmesni čakalni vrsti in b) da se bo pokvaril zaradi bitnih napak v fizičnem prenosnem kanalu [4]. Namen naloge je bil opazovati rezultate simulacij TCP v primeru prenosa po neki poenostavljeni zvezi. To je lahko zveza v fiksnem omrežju ali brezžični kanal, lahko je kombinacija obeh. Glavno merilo pri primerjanju je učinkovitost protokola. Učinkovitost protokola med izvajanjem simulacije je definirana kot kvocient med časom, potrebnim za oddajo vnaprej določenega števila segmentov, predvidenega v simulaciji, in dejanskim časom izvajanja simulacije. Čas, potreben za prenos segmentov, je produkt med njihovim številom in časom oddaje enega segmenta. Čas izvajanja simulacije narašča v izgubnih okoljih zaradi ponovnih oddaj segmentov. Čas izvajanja simulacije narašča, zato ker mora oddajnik izgubljene segmente ponovno oddati in po potrebi zmanjšati hitrost oddajanja [7]. Vrednosti učinkovitosti so opazovane pri različnih vrednostih parametrov, kot so širina drsečega okna, dolžina prenašanih segmentov in dolžina čakalne vrste. Učinkovitost je odvisna od lastnosti simuliranega protokola. Neodvisna parametra, ki ju spreminjamo, sta pogostost oddajanja bremenskih segmentov, namenjenih simulaciji zamašitev v vmesni čakalni vrsti, in pogostost bitnih napak. Simulacija je izvedena na več načinov. Simulirane izgube so lahko posledica zamašitev v vmesni čakalni vrsti, ali so posledica bitnih napak v prenosnem kanalu. Tretja možnost so izgube zaradi zamašitev in bitnih napak istočasno. Temu ustrezno je definiran simulacijski model v načinih 1, 2 in 4. Namen simulacije v načinu 1 je pokazati, kako se protokol TCP odziva na izgube zaradi zamašitev. Segmenti, ki pridejo v čakalno vrsto, so lahko zakasnjeni ali zavrženi, kar je odvisno od števila segmentov, ki so že v čakalni vrsti. V stacionarnih internetnih omrežjih so mogoče zamašitve in TCP je narejen za takšna omrežja. Način 2 predstavlja simulacijo prenosa po fizičnem kanalu z bitnimi napakami. Primer takšnega kanala je brezžični kanal. Sprejemnik označi segmente z napakami kot pokvarjene. Pokvarjeni segmenti so zavrženi. Oddajnik ponovno pošlje segmente, ki so imeli napake in so bili zavrženi. Način 4 predstavlja simulacijo prenosa z zamašitvenimi izgubami in bitnimi napakami istočasno. Namen takšne simulacije je prikazati, kako se TCP obnaša v primeru prenosa v kombiniranih omrežjih. V nalogi je prikazan tudi poenostavljen protokol TCP, prilagojen za prenos po brezžičnem kanalu, imenovan WTCP (wireless trasnmission control protocol). Rezultati simulacij TCP in WTCP v načinu 2 so primerjani z namenom ugotavljanja učinkovitosti poenostavljenega in prilagojenega WTCP. WTCP ne uporablja mehanizmov za krmiljenje zamašitev, kot so počasni zagon, izogibanje zamašitvam, hitra ponovna oddaja po treh ponovljenih potrditvah in hitro okrevanje. Odddajnik že na podlagi prve ponovljene potrditve ponovno pošlje segment, za katerega pričakuje, da je bil izgubljen na prenosni poti. Taka ponovna oddaja se za določen segment izvaja enkrat. V primeru več izgub ostane kot reševalni mehanizem iztek časovnika T za ponovno oddajo. Protokol TCP v kombiniranih omrežjih naj bi imel dodane mehanizme za ločevanje različnih vrst izgub, saj nepotreben zagon mehanizmov za krmiljenje zamašitev upočasni prenos podatkov. Krmiljenje zamašitev je potrebno, ko so segmenti izgubljeni v vmesnih čakalnih vrstah zaradi zamašitev. V primeru bitnih napak v prenosnem kanalu in posledično zavrženih segmentov, ki so bili pokvarjeni, so ti nepotrebni in je ena izmed rešitev poseben protokol na brezžičnem delu povezave, kot je npr. WTCP, ki ne uporablja zamašitvenih mehanizmov. Ena izmed rešitev pa bi bila tudi spremenjen protokol TCP.

• Language: Slovenian
• Keywords: TCP, simulacija, zamašitev, bitne napake, izkoristek
• Work type: Master's thesis
• Organization: FE - Faculty of Electrical Engineering
• Year: 2016
• PID: 20.500.12556/RUL-84121
• Publication date in RUL: 11.07.2016

Secondary language

• Language: English
• Title: TCP PROTOCOL EFFICIENCY IN DIFFERENT NETWORK ENVIRONMENTS

Abstract

The theme of the task is the simulation of TCP in different network environments. Specification of the protocol TCP (transmission control protocol) in the basic mode of operation without losses is given in the formal programming language SDL (specification and description language). Then is carried out the transition in the programming code in the programming language C. The code in the C presents the simulation program together with a simulator, which contains a variety of simulation functions, among them random allows the implementation of experiments and an insight into the effectiveness of the TCP protocol when it comes to different network environment with losses. Random functions determine a) the probability that some segment in the transmission lost due to clogging in a random generated load transport in the intermediate queue, and b) that will ruin due to bit errors in the physical transmission channel [4]. The purpose of the tasks was to observe the results of the simulations of TCP in the case of a transfer by a simplified connection. This can be a connection in the fixed network or the wireless channel, it can be a combination of both. The main criterion when comparing the performance is the effectiveness of the protocol. The efficiency of the protocol during the implementation of the simulation is defined as ratio between the time necessary for transmitting of a predetermined number of segments projected in the simulation and the actual time of the implementation of the simulation. The time required for the transfer of segments is the product between their number and the time of the transmitting of one segment. The time of simulation is increasing in the loss environments due to the re-transmitted segments. The duration of the simulation increases because the transmitter must retransmit lost segments and where appropriate to reduce the speed of transmitting [7]. The values of efficiency are observed for different values of the parameters such as the width of the sliding window, the length of the segments and queue length. Efficiency depends on the properties of the simulated protocol. The independent parameters that we change are the frequency of transmitting load segments intended for simulation of clogging in the intermediate queue and the frequency of the bit errors. Simulation is carried out in several ways. The simulated loss can be a result of blockage in the intermediate queue, or are the result of bit errors in the transmission channel. The third option is the loss due to clogging and bit errors at the same time. Accordingly, it is a defined simulation model in modes 1, 2 and 4. The purpose of the simulation in mode 1 is to show how TCP reacts to losses due to blockage. The segments that arrive in the queue can be delayed or discarded, depending on the number of segments that are already located in the queue. In the fixed internet networks are possible congestion and the TCP is made for such networks. Mode 2 means the simulation of the transmission by the physical channel with bit errors. An example of such a channel is the wireless channel. Receiver mark segments with errors as corrupt. Corrupted segments are discarded. The transmitter re-sends the segments that had errors and were discarded. Way 4 means the simulation of the transmission losses due to the congestion and bit errors at the same time. The purpose of such simulations is to show how TCP behaves in case of a transfer in combined networks. In the task is also shown simplified TCP adapted for transmission through a wireless channel, called the WTCP (wireless trasnmission control protocol). The results of the simulations for the TCP and WTCP in mode 2 are compared with the aim to determine the effectiveness of a simplified and adjusted protocol WTCP. WTCP does not use mechanisms to control congestion, such as slow start, avoiding congestion, fast re-sending after three repeated confirmation and fast recovery. Odddajnik already on the basis of the first repeated confirmation re-sends the segment, for which it is expected that has been lost on the transmission path. Such resending the segment carried out only once. In the case of multiple losses remains as a rescue mechanism expiry timer T for resending. Protocol TCP in combined networks must have an mechanism for the separation of different types of losses. Mechanisms for the control of the congestion will slow down the data transfer. TCP should used them when segments were lost in the intermediate queues due to congestion. In the case of bit errors in the transmission channel and consequently the discarded segments that were corrupt is one of the solutions a special protocol on the wireless part of the connection such as WTCP. One of the solutions, it would also be a modified TCP.

• Keywords: efficinecy, TCP, simulation, congestion, bit errors