Development of a Low-Cost Link Training and Auto-Negotiation Sniffer for 10G SerDes

This thesis explores the world of high-speed SerDes communication systems, focusing on the methodologies for link training and equalization techniques. As Ethernet has evolved, increasing data transmission rate introduces new challenges for the physical link. To address these challenges, traditional Ethernet sniffers or protocol analyzers, despite their high cost, have been the tools of choice for debugging and testing network traffic. This thesis proposes a cost-effective alternative by Building and characterizing a system able to sniff the data and debug it between two peers using an oscilloscope and RF components. Utilizing an RF power divider as a pick-off, this innovative approach allows for the capture, analysis, and debugging of network traffic. The introductory chapter outlines the context, objectives, and challenges addressed by the thesis. The subsequent literature review provides an overview of modern transceivers (SerDes), explores the concept of equalization with a focus on mitigating inter-symbol interference (ISI), and analyzes high-speed SerDes communication techniques such as NRZ and PAM4 modulation. The third chapter delves into the process of link training, including auto-negotiation and the structure of link training frames, emphasizing the importance of control channels according IEEE Standards 802.3 . In the fourth chapter, the experimental setup is detailed, including a block diagram of key components, experimental procedures, and conducted experiments. The thesis in the last chapter concludes by summarizing its findings, notably the successful sniffing and debugging of link training signals, as well as the open issues and the future prospects.