TSDL v2.1

Invented by Giovanni Blu Mitolo
Originally published: 20/11/2017, latest revision: 9/11/2018
Related implementation: /src/strategies/ThroughSerial/
Compliant versions: PJON v10.0 and following
Released into the public domain

TSDL (Tardy Serial Data Link) is a simplex or half-duplex serial data link that supports both master-slave and multi-master modes. It supports collision avoidance, reliable frame separation and optional synchronous response to frame transmissions.

 ______ TX   RX ______
|      |-------|      |
        RX   TX

TSDL can be used to establish a point-to-point link between two devices if used with a bare serial link, or to support one or many to many communication using radio or RS485 transceivers.

Medium access control

TSDL operates in master-slave mode applying the request-response procedure. Being impossible to implement a carrier-sense procedure over a serial port TSDL uses variation of slotted ALOHA medium access method for multi-master mode. Before a frame transmission the serial buffer is read, if not empty ongoing communication is detected and collision avoided, if empty for a duration longer than the response time-out plus a short random time, frame transmission starts in which the packet is entirely transmitted. Of all contention based random multiple access methods, slotted ALOHA, which maximum data throughput is only 36.8% of the available bandwidth, is one of the least efficient, therefore TSDL in multi-master mode should not be used in networks where many devices often need to arbitrarily transmit data.

Frame transmission

Before a frame transmission the communication medium is analysed, if any data is received communication is detected and collision is avoided, if logic 0 is detected for a duration longer than the response time-out plus a small random time, data is transmitted encapsulated in a SFSP (Secure Frame Separation Protocol) v1.0 frame.

Synchronous response

A frame transmission in both master-slave and multi-master modes can be optionally followed by a synchronous response of its recipient, all devices must use the same response time-out to avoid collisions. The acknowledgment reception phase must be shorter than the response time-out to be successful.

Transmission                                    Response
 _______  ______  ______  _____                   _____
|  149  ||  H   ||  I   || 234 | LATENCY         |  6  |
|_______||______||______||_____|                 |_____|

The required response time-out for a given application can be determined practically transmitting the longest supported frame with the farthest physical distance between the two devices. The highest interval between packet transmission and acknowledgement measured plus a small margin is the correct time-out that should exclude acknowledgement losses.

This document is automatically generated from the github repository. If you have noticed an error or an inconsistency, please report it opening an issue here