MASTER THESIS: CANtegrity: CAN FD CAN-bit validation test

Originally, CAN was a computer network using low bit rate where Signal Integrity was not a problem because the signal’s symbol time was low compared to signal delay in the network. The arbitration rule demanded the bit-length to be almost 4 times the signal delay over the network. This limitation, implied by the arbitration rule, resulted in bit rates low enough to prevent the occurrence of signal integrity problems, even if impedances were not well matched. With the introduction of CAN-FD it became possible to increase the bitrate, thereby becoming more sensitive to impedance variation. The configuration of the CAN-bit, which is important also for CAN-CC (Can Classic), became even more important when using the complex bitrate switch in CAN FD. The first problem is to secure that all installed units in the network have a correct CAN-bit configuration. This is complex because the setting depends not only on the units themselves but also on the clock variations and on the signal delay to all other installed units. Added to this, it is necessary to understand the signal integrity problem caused by the higher bitrate. By having full control of the configuration in every installed unit, it is possible to reduce the problem to a minimum. 

This master thesis is about developing a test function that can be used to determine the CAN-bit parameters for a CAN network {TQ-length (ns), number of TQ in the CAN-bit, number of TQ to the Sample point, number if TQ in the SJW}. 

The first task is to develop an FPGA logic in Verilog that generates a specific test pattern. This is to be included in the CANtegrity CAN-FD IP that is used in Kvaser’s hardware. In the test, the signal resulting from the generated test pattern will be recorded by test software that based on the recorded data calculates the CAN-bit parameters in the unit during the performed test. This test is carried out with only one unit installed. The main part of the thesis work is the software that searches for pattern in the recording. This is not time critical so the software can be some sort of high-level language (Python). 

If it fits into the time frame, the next task will be to do the same using two units connected to the common CAN-bus. The problem here is that both connected units could respond concurrently which can make the calculation or analysis more complicated. 

The ultimate solution is to run the test on a CAN-bus with N nodes. It is obvious that it is necessary to first complete the test for N=1 and N=2 before this part of the task is started. In this case it is expected to also measure the distance (in nanoseconds) to all installed nodes. Kvaser already has a software that can fingerprint every CAN-frame and it is thereby possible to find out how many different transmitters that are connected to the CAN-bus. This knowledge makes it possible to focus the analysis on each transmitter and based on the transmitting CAN controller behaviour obtain the distance to the unit.

Author: Kent Lennartsson

Generic scenario simulator with a Digital Twin approach 

A master thesis for two students Radio devices for communication are often exposed to highly dynamic environments, whether it concerns the systems employing the radio devices or the physical environment in which the radio devices move, thereby causing significant changes in propagation conditions or changed levels of interference from various emitters and noise sources. 

Moreover, radio links may be part of a system comprising multiple radio devices which implies a complex interrelation between the radio devices and hence very complex scenarios, difficult to test, especially repetitively. The latter would be desirable to enable trouble shooting and studying of specific situations to draw valuable conclusions from the behaviour of the radio system. Computer based simulations have the advantages of speeding up the process of validation, at the same time providing statistical confidence. 

To cope with dynamic environments, a radio system needs to have dynamic properties too, whether realized as rules based algorithms or simpler regulating control processes. The dynamic properties could e.g. be about optimizing the link capacity in different directions, changing modulation and transmit power or adapting the set of employed frequencies. The radio devices’ abilities to become aware of this dynamic environment may be possible to different degrees, from simple link quality estimation and sensing of available frequencies to more advanced methods where the signals are analysed in various ways. 

This thesis work is about exploring models for a generic scenario simulator that can be used to try out various dynamic algorithms to be employed in certain given scenarios. The scenarios may be based on recorded data exchanges from real physical set-ups or on assumptions of their behaviour. Scenarios may also be based on tracked or assumed movements of radio devices in an area for which the link qualities will be established by means of special analysis tools not part of the thesis work. This includes different means to represent the awareness of the dynamic environment. 

The scenario simulator should be designed as a “digital twin” to an existing physical simulator consisting of up to seven pairs of radio devices and two sources of interference in which the radio propagation conditions and the load with respect to generated messages can be dynamically changed. The scenario simulator shall implement the dynamic algorithms autonomously acting on various stimuli, a digital representation of the awareness information to be the outcome from a transformation of information provided by the analysis of real physical scenarios.

The scenario simulator provides the ability to alter the dynamic algorithms to a great extent and to design advanced behavioural models for autonomous acting. These algorithms should allow for incorporation into the control logic of the radio devices to enable validation in the physical scenario simulator (a test rig). This requires a time base implementation that allows adaptation to real time operative systems.

Author: Anders Bäckman

Multidrop network analysis and simulation 

The goal is to have a program that will help users to understand what you can do with a multidrop network without jeopardizing the signal integrity. When bit-rate increases, it is necessary to consider the signal integrity of the cable layout. There are a lot of tools and books covering Radio Frequency (RF) signals for a narrow band signal that can be found. It is however a little more complicated for digital data communication where the energy is spread over a several frequencies. 

Even for digital signals there are several books on the subject and instruments to analyze the problem. There are several suppliers of instruments working both in the frequency and time domain that handle this problem, but they assume a Point-to-Point transmission line at a relatively high frequency. No books exist that discuss the complex multidrop communication used in CAN, 10BASE-T1S or 802.3dg. The frequency range is relatively low (below200 MHz) but the transmission line is a very complex network with multiple drop lines to the transceiver circuits. For such complex networks it is not possible to use TDR (Time Domain Reflectometer) and VNA (Vector Network Analyzer) in the same way as they are used for a Point-to-Point network. 

This problem could be solved by the generic program ADS (Advance Design Systems) provided by Keysight which is designed to solve any problem involving both high bandwidth and frequency. This is a complex and expensive tool. What we need is a program that solves only our specific problem that is relatively simple but involves a lot of calculation. 

Handling this task will demand knowledge about S-parameters and programming (Python) to mathematically process the S-parameters. The suggestion is to obtain adequate understanding of the problem by carefully studying the information found in the book “S-parameters for Signal Integrity” by Peter J. Pupalaikis ISBN 978-1-108-48996-6. This book has reference to a software package (SignalIntegrity) written in Python that it is suggested to be used as the platform for this program. 

Instead of modelling the CAN/802.3dg multidrop networks for simulation in Spice, it is suggested to build the layout with S-parameter blocks to describe the multi-drop network. This will make it possible analyze the system’s expected signal integrity without having to know every minute detail. It is thereby possible to define the allowed S-parameters that a node can have and use this as boundaries for the S-parameters of the nodes to be installed. 

It can also be possible to measure the S-parameters for the installed nodes and use that as input to check if the system can provide the necessary signal integrity. 

Author: Kent Lennartsson

Scalable two-layered computer architecture 

A master thesis for two students 

Layering of communication protocols is an important pre-requisite for standardised communication systems. When it comes to radio technology, this may be considered even more important. This is because changes to radio functionality may require very costly re-certification of a radio device. With a layered approach, it could be possible to isolate its radio functions to a degree that ensures that the impact of changes to the radio device do not impact the performance and behaviour from a certification aspect. Also, the extent of regression tests and verification is reduced if it is possible to limit any change to a certain layer. 

The desired situation would be to contain the software related to the radio functions in a separate processing unit. The radio functions would thereby be regarded as hierarchically lower and closer to the physical layer which is sometimes referred to as Media Access layer (MAC). The radio functions would thereby be accessible via a clearly defined interface to a radio transceiver and a suitable Application Protocol Interface (API). This layered approach also provides scalability in the sense that it would be possible with multiple radio transceivers connected to one radio controller. 

This thesis work is about exploring methods to create a layered architecture for a radio solution which could be a combined radio controller and radio transceiver or a distributed more complex constellation with multiple radio transceivers, all connected to the same radio controller which runs multiple processes in a co-ordinated way. This co-ordination would be of interest in solutions where multiple transceivers cannot be allowed to transmit simultaneously. Hence, the characteristics of the system, particularly that of the mentioned interface, is highly important when it comes to synchronization of multiple radio transceivers and their processing units. Latency is another important property of the interface with its protocol in that it will have a direct impact on the delay characteristics of a radio connection, especially in cases where the radio controller interfaces to multiple radio transceivers that are to be employed simultaneously. 

The layered architecture shall be implemented in a set-up consisting of a processor board representing the radio controller and another processor board that represents the radio functionality. The latter should be connected to a radio transceiver. A suitable interface (including API) shall be implemented to convey control information between the radio controller and the processor of the radio transceiver which is emulating the radio chip. This set-up should be modifiable in such a way that the emulation can be replaced by a connection to a radio chip, thereby enabling the set-up to communicate with an existing radio device based on the traditional single processor architecture. 

The thesis work includes separating an existing but limited software into a two layered architecture which is supposed to be interoperable with the existing software. 

Author: Anders Bäckman

Time synchronization by gPTP 

The NTP (Network Time Protocol) is well known in the LAN (Local Area Network) and internet connections. For a small LAN, its precision varies from a few milliseconds down to a 1 millisecond. For Ethernet there is standard IEEE1588 which is the basis for PTP (Precision Time Protocol) that provides a sub microsecond precision. The problem with this standard is the assumption that the latency is constant over time and symmetric. Moreover, it is not an integrated part in IEEE802 (Ethernet) standard. This has been solved in the standard IEEE802.1AS, named gPTP (generalized Precision Time Protocol) that requires special Ethernet hardware but that provides good precision over time independently of hardware supplier. The IEEE802.1AS standard is also the time base for TSN (Time Sensitive Networking) which is a hardware function that with several different means secures the communication latency, that the communication is not exhausted, and by that ensures that the most important information has priority. 

When logging data, it is important to have the same time base for all sources of data. For our USB-interfaces that problem is solved with Kvaser’s MagiSync.technology. 

Some products are based on Ethernet where MagiSync is not directly applicable. For that reason, it is of interest to understand how it would be possible to use gPTP in Kvaser’s products. 

The goal of the master thesis is to develop software for hardware to be selected that in combination with the hardware provide functions as described IEEE802.1AS. 

The work will start by studying the IEEE802.1AS to understand the function and which hardware will be required to obtain a working solution. It will be necessary to find an already existing hardware because there will be no time to design and produce any specific hardware. As a worst case, it could be necessary to design an FPGA logic that can provide the necessary hardware functionality to obtain a working solution. 

The thesis work requires some knowledge about software preferably RUST or C. Should no suitable hardware be found it would be necessary with knowledge about Verilog that is used to emulate hardware. 

Author: Kent Lennartsson