LinkBo: An Adaptive Single-Wire, Low-Latency, and Fault-Tolerant Communications Interface for Variable-Distance Chip-to-Chip Systems

Bochen Ye 1 , 3 , Gustavo Naspolini 2 , Kimmo Salo 2 , Manil Dev Gomony 1 · 2025

Read LinkBo: An Adaptive Single-Wire, Low-Latency, and Fault-Tolerant Communications Interface for Variable-Distance Chip-to-Chip Systems by Bochen Ye 1 , 3 , Gustavo Naspolini 2 , Kimmo Salo 2 , Manil Dev Gomony 1 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-01339.txt.

Introduction

Cheap embedded systems often talk chip to chip over a single wire, saving pins, but older one-wire protocols are slow and fragile. Bochen Ye, Gustavo Naspolini, Kimmo Salo, and Manil Dev Gomony present LinkBo, a single-wire protocol built for low delay, higher throughput, and fault tolerance over varied distances. It adds a hardware interrupt so urgent messages are not stuck waiting.

At the protocol level, high-priority messages with error detection arrive in about fifty microseconds, at least twenty times faster than the commercial 1-Wire option and several times faster than UNI/O. The team built the hardware design and tested it on two FPGA boards. LinkBo held up over wires as long as fifteen meters at three hundred kilobits per second, and reached seven and a half megabits per second over a short eleven centimeter wire.

Catalog record

CREATORBochen Ye 1 , 3 , Gustavo Naspolini 2 , Kimmo Salo 2 , Manil Dev Gomony 1
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-01339

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

LinkBo: An Adaptive Single-Wire, Low-Latency, and Fault-Tolerant Communications Interface for Variable-Distance Chip-to-Chip Systems Bochen Ye 1 , 3 , Gustavo Naspolini 2 , Kimmo Salo 2 , Manil Dev Gomony 1 Abstract Cost-effective embedded systems necessitate utilizing the single-wire communication protocol for inter-chip communication, thanks to its reduced pin count in comparison to the multi-wire I2C or SPI protocols. However, current single-wire protocols suffer from increased latency, restricted throughput, and lack of robustness. This paper presents LinkBo, an innovative single-wire protocol that offers reduced latency, enhanced throughput, and greater robustness with hardware-interrupt for variable-distance inter-chip communication. The LinkBo protocol-level guarantees that high-priority messages are delivered with an error detection feature in just 50.4 µs, surpassing current commercial options, 1-wire and UNI/O by at least 20X and 6.3X, respectively. In addition, we present the hardware architecture for this new protocol and its performance evaluation on a hardware platform consisting of two FPGAs. Our findings demonstrate that the protocol reliably supports wire lengths up to 15 meters with a data rate of 300 kbps, while reaching a maximum data rate of 7.5 Mbps over an 11 cm wire, providing reliable performance for varying inter-chip communication distances. [email protected], {gustavo.naspolini,kimmo.salo}@nxp.com, [email protected] I Introduction Common chip-to-chip communication methods typically rely on protocols such as Serial Peripheral Interface (SPI) or Inter-Integrated Circuit (I2C) [ 1 ] . SPI uses four wires: one for the clock signal, one for chip select, and two for data transmission. In contrast, I2C streamline communication with just two wires: one for data and one for the clock signal. However, as the number of interconnected chips increases, the total number of required pins for both SPI and I2C grows accordingly. In Fig. 1 (left), the total number of communication pins required for SPI increases sharply while I2C exhibits a more linear growth. In embedded systems, minimizing the physical footprint, particularly the number of off-chip communication wires, is a critical design goal. Reducing the number of external pins can significantly shrink the chip’s packaging area, potentially saving several square centimeters. According to the packaging information [ 2 , 3 ] , a 16-pin package occupies only 40% of the area of a 20-pin package and incurs just 60% of the cost (Fig. 1 ). These findings highlight the strong correlation between pin count and both packaging area and cost. Consequently, reducing the number of external pins is essential for area-constrained applications. Although SPI and I²C require only 2 to 4 communication pins, they may still be suboptimal for systems with stringent area or cost limitations. (a) Figure A (b) Figure B Fig. 1: The total communication pin-count cost with different communication protocol in multi devices system (left). The area and package cost between 20-pin package and 16-pin package (right). A single-wire communication protocol offers a promising alternative by single pin design. ”Single-wire” refers to a class of protocol that uses a single wire for communication. The state-of-the-art (SOTA) single-wire protocol, known as the 1-wire protocol, operates as a half-duplex serial bus using only one data line and a single pin [ 4 ] . However, the SOTA 1-wire protocol prioritizes long-distance communication, leading to extended bit-slot durations, which result in significant message latency and a low bit rate of 16.3 kbps, failing to meet the high-speed processing requirements of modern applications. Furthermore, 1-wire communication protocol is not particularly robust as it becomes susceptible to failure if the host malfunctions, and it lacks mechanisms to secure data integrity. To address these issues, this paper proposes the following contributions: 1. We propose LinkBo protocol, which is a novel chip-to-chip single-wire protocol that can achieve lower latency, higher throughput, and improved robustness. It also contain a system-level model to analysis performance. 2. We propose a novel hardware architecture for different speed and variable-distance LinkBo protocol on two FPGAs with only single wire. 3. Performance comparison of LinkBo with SOTA single-wire protocols in terms of latency and bit rate. In addition, the performance of the LinkBo protocol is analyzed under different parasitic parameters, wire lengths, and bit rate (clock frequencies) to assess robustness. The rest of the paper is organized as follows. Section II introduces the foundational concepts and examines related work on the single-wire protocol. Section III outlines the functionality of the Linkbo protocol. Section IV offers a high-level overview of the channel model. Section V details the hardware architecture of our transmitter, receiver, and driver. Section VI presents the hardware implementation on FPGA and verifies its correctness. Section VII presents the FPGA setup and experimental results with respect to bit rate, latency, physical parameters, wire length, and clock frequency. Finally, the paper concludes with a summary of our findings and suggestions for future research in Section VIII . II Background and Related work This section begins with an introduction to the current single-wire protocol, followed by a summary of related work. II-A 1-wire protocol overview The 1-wire bus [ 4 , 5 ] , which is a low-speed single-wire serial protocol designed for long-distance wired communication (extending up to 100 meters), is depicted in Fig. 2 . This bus architecture features a single host that exerts control over multiple slave devices. As illustrated in [ 5 ] , the communication process is initiated by the host transmitting a 960 µs reset pulse to all connected slaves. This is succeeded by a 64-bit device select message intended to address a specific slave, with each bit slot having a duration of 70 µs. Subsequently, the host may transmit either a 64-bit data message or a 16-bit command. In order to establish communication with an alternative slave, the host is required to recommence the reset pulse procedure. Communication between two chips is typically asynchronous due to different clock frequencies or unaligned clock edges, which can lead to errors at the receiver.To tackle this problem, it is essential to use either encoding techniques or clock synchronization strategies. Fig. 2: Current 1-wire bus architecture. The host contain processing core, function units, memory and 1-wire protocol. The host use 1-bit wire to send message to multiple slaves and receive data from slaves. II-B Asynchronous communication The conventional 1-wire protocol uses low voltage to represent a 0 bit and high voltage to represent a 1 bit. This method introduces a DC component, which makes clock synchronization difficult and increases susceptibility to interference. In contrast, Manchester encoding eliminates the DC component, is self-clocking, and offers strong resistance to interference because each Manchester bit slot contains a transition. This transition can be utilized for both error detection and clock synchronization [ 6 ] . Fig. 3: Different types of encoding method. Current 1-wire uses NRZ (Non-return-to-zero) code. IEEE802.3 standard of Manchester is just the opposite of the G.E.Thomas standard. There are two standards for Manchester code: IEEE802.3 used in Ethernet and token bus network, the G.E.Thomas standard, which is the original version of Manchester encoding. Detail are shown in Fig. 3 . IEEE802.3 uses a rising edge to represent a 1 bit and a falling edge to represent a 0 bit, while the G.E.Thomas standard uses the opposite conventions. In this thesis, we use IEEE802.3 as our Manchester encoding standard, and the encoding method is illustrated in Fig. 4 . Fig. 4: Encode method

Read the complete text