Number of Visitors: 1508
 

:

M.Sc. Thesis Defense Session

Introducing a structured data transfer protocol for IoT limited-bandwidth dynamic computational environments

News Code: 14

Publishing Date: 1 Sep 2022 4:8



In The Name of God

M.Sc. Thesis Defense Session

‚Äč

Computer Architecture Engineering Group

 

Supervisor:

Dr. Ali Bohlooli

Advisor:

Dr. Kamal Jamshidi

Internal Reviewer:

Dr. Mohammad Reza Khayyambashi

External Reviewer:

Ahmadreza Montazerolghaem

Researcher:

Raffi Dilanchian

Date: 7 September 2022

Time: 8:30 AM

Location:

Ansari building, Third floor, Dr. Braani Hall

 

Topic:

Introducing a structured data transfer protocol for IoT limited-bandwidth dynamic computational environments

Most IoT devices operating in mobile computing environments have networks with variable error rates and limited bandwidth. The mechanism of resending duplicate packets to recover lost information during transmission, which is used in most of the common data transmission protocols, causes excessive bandwidth usage and increases transmission delay. One of the methods to solve this problem is the usage of forward error correction approaches such as random linear network coding mechanism in information transmission. The main problem with most of the RLNC-based protocols is that the coding ratio does not change during data transmission according to the network conditions. In this research, with the aim of solving this problem, a new information transfer protocol based on the block-based RLNC approach with systematic data coding called ARLNC has been introduced, in which the coding ratio and sending information amount are adjusted dynamically based on the estimated network error rate in run-time. In the proposed ARLNC approach, the network error rate is estimated with the help of the receiver feedback values. The encoding and decoding calculations of this method were performed in the GF(28) Galois field, and the Gauss-Jordan Elimination algorithm was used in the packet decoding process. ARLNC approach has been implemented with the help of Python programming language and its performance has been simulated by the discrete event method for erasure with errors, Gilbert Elliott, exponential and constant rate error models. The findings of this research indicate that by adjusting the encoding rate and the amount of data sent according to the changes in the network error rate, increases the throughput of the network and reduces the data transmission delay in most cases. The results of the comparisons made in this research show that the transmission delay in the proposed approach is on average one third and the throughput is three times that of the standard RLNC approach with a fixed coding rate. Also the computational cost of the ARLNC approach is on average 68% higher than the standard RLNC approach.