For the last few decades, we have witnessed an ever-growing expansion of wireless networks and applications. Although these networks are subject to transmission errors, we manage to have reliable data delivery through the use of protocols such as the universally used Transmission Control Protocol (TCP). The latter uses retransmissions, when the received data is corrupted, to ensure reliable communications at the expense of an increase in transmission delays and overhead. To avoid these drawbacks, improved methods for error correction and packet recovery are needed.
Many approaches have been introduced to cope more efficiently with the vulnerability of wireless networks to transmission errors. But they all come with specific advantages and disadvantages. The Partial Packet Recovery (PPR) is and among the most widespread and promising ones. It utilizes the partially received packets to recover the erroneous packet and thus attempts to avoid retransmissions as much as possible.
In this thesis, we aim at correcting and recovering corrupted packets at the receiver instead of ignoring them and asking for retransmissions. Accordingly, we introduce three methods at the receiver’s side : Majority Voting with Checksum Validation (MVCV), Dissimilarity-based Enumeration Packet Recovery with Checksum Validation (DEPRCV), and Cross-Layer Packet Recovery (CLPR). They all attempt to reconstruct the error-free packets relying on one or several erroneously received copies. MVCV relies on the majority of bit value occurrences among received packets at each bit position. DEPRCV identifies bit positions where different values are observed within each received packet and constructs a list of possible error-free packets. In both methods, the candidate reconstructed packets are validated with the Transmission Control Protocol (TCP) checksum. Finally, CLPR combines a recently proposed CRC-based Error Correction (CRC-EC) method with DEPRCV and MVCV. A key benefit of these methods is that it doesn’t require any change to TCP, i.e., compatible with TCP, and can be easily implemented in a client.
Simulation results show the superiority of the proposed methods in terms of the number of retransmissions and transmission delay when compared to TCP and CRC-based Error Correction (CRC-EC). They also improve effective network throughput and energy efficiency. For instance, at a Bit Error Rate (BER) of 0.0001, TCP and CRC-EC both require 3 retransmissions to deliver reliably 99.9% of the packets while MVCV, DEPRCV and CLPR require 2, 2, and 1 retransmissions, respectively, in order to reach the same 99.9% reliability level. For a higher BER of 0.001, for the reliable delivery of 99.9% of the packets, TCP and CRC-EC require at least 20 and 8 retransmissions, respectively, while MVCV, DEPRCV and CLPR require 3, 1, and 1 retransmissions.
| Date | 26 Jul 2021 |
|---|
| Original language | American English |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Stéphane Coulombe (Supervisor) |
|---|
Naghashi, M. (Author),
Coulombe (Supervisor),
26 Jul 2021Student thesis: Master's thesis › Master in Engineering: Information Technology Engineering