Advanced

Virtual Cycle-accurate Hardware and Software Co-simulation Platform for Cellular IoT

Tovar Rascon, Marcel LU and Elfborg, Patrik LU (2017) EITM01 20162
Department of Electrical and Information Technology
Abstract
Modern embedded development flows often depend on FPGA board usage for pre-ASIC system verification. The purpose of this project is to instead explore the usage of Electronic System Level (ESL) hardware-software co-simulation through the usage of ARM SoC Designer tool to create a virtual prototype of a cellular IoT modem and thereafter compare the benefits of including such a methodology into the early development cycle. The virtual system is completely developed and executed on a host computer, without the requirement of additional hardware. The virtual prototype hardware is based on C++ ARM verified cycle-accurate models generated from RTL hardware descriptions, High-level synthesis (HLS) pre-synthesis SystemC HW accelerator models and... (More)
Modern embedded development flows often depend on FPGA board usage for pre-ASIC system verification. The purpose of this project is to instead explore the usage of Electronic System Level (ESL) hardware-software co-simulation through the usage of ARM SoC Designer tool to create a virtual prototype of a cellular IoT modem and thereafter compare the benefits of including such a methodology into the early development cycle. The virtual system is completely developed and executed on a host computer, without the requirement of additional hardware. The virtual prototype hardware is based on C++ ARM verified cycle-accurate models generated from RTL hardware descriptions, High-level synthesis (HLS) pre-synthesis SystemC HW accelerator models and behavioural models which implement the ARM Cycle-accurate Simulation Interface (CASI). The micro-controller of the virtual system which is based on an ARM Cortex-M processor, is capable of executing instructions from a memory module.

This report documents the virtual prototype implementation and compares both the software performance and cycle-accuracy of various virtual micro-controller configurations to a commercial reference development board. By altering factors such as memory latencies and bus interconnect subsystem arbitration in co-simulations, the software cycle-count performance of the development board was shown possible to reproduce within a 5% error margin, at the cost of approximately 266 times slower execution speed. Furthermore, the validity of two HLS pre-synthesis hardware models is investigated and proven to be functionally accurate within three clock cycles of individual block latency compared to post-synthesis FPGA synthesized implementations.

The final virtual prototype system consisted of the micro-controller and two cellular IoT hardware accelerators. The system runs a FreeRTOS 9.0.0 port, executing a multi-threaded program at an average clock cycle simulation frequency of 10.6 kHz. (Less)
Popular Abstract
-Designing and simulating embedded computer systems virtually.

Cellular internet of things (IoT) is a new technology that will enable the interconnection of everything: from street lights and parking meters to your gas or water meter at home, wireless cellular networks will allow information to be shared between devices. However, in order for these systems to provide any useful data, they need to include a computer chip with a system to manage the communication itself, enabling the connection to a cellular network and the actual transmission and reception of data. Such a chip is called an embedded chip or system.

Traditionally, the design and verification of digital embedded systems, that is to say a system which has both hardware... (More)
-Designing and simulating embedded computer systems virtually.

Cellular internet of things (IoT) is a new technology that will enable the interconnection of everything: from street lights and parking meters to your gas or water meter at home, wireless cellular networks will allow information to be shared between devices. However, in order for these systems to provide any useful data, they need to include a computer chip with a system to manage the communication itself, enabling the connection to a cellular network and the actual transmission and reception of data. Such a chip is called an embedded chip or system.

Traditionally, the design and verification of digital embedded systems, that is to say a system which has both hardware and software components, had to be done in two steps. The first step consists of designing all the hardware, testing it, integrating it and producing it physically on silicon in order to verify the intended functionality of all the components. The second step thus consists of taking the hardware that has been developed and designing the software: a program which will have to execute in complete compliance to the hardware that has been previously developed. This poses two main issues: the software engineers cannot begin their work properly until the hardware is finished, which makes the process very long, and the fact that the hardware has been printed on silicon greatly restricts the possibility of doing changes to accommodate late system requirement alterations; which is quite likely for a tailor-made application specific system such as a cellular IoT chip.

A currently widespread technology used to mitigate the previously mentioned negative aspects of embedded design, is the employment of field-programmable gate array (FPGA) development boards which often contain a micro-controller (with a processor and some memories), and a gate array connected to it. The FPGA part consists of a lattice of digital logic gates which can be programmed to interconnect and represent the functionality of the hardware being designed. The processor can thus execute software instructions placed on the memories and the hardware being developed can be programmed into the gate array in order to integrate and verify a full hardware and software system. Nevertheless, this boards are expensive and limit the design to the hardware components available commercially in the different off-the-shelf models, e.g. a specific processor which might not be the desired one.

Now imagine there is a way to design hardware components such as processors in the traditional way, however once the hardware has been implemented it can be integrated together with software without the need of printing a physical silicon chip specifically for this purpose. That would be extremely convenient and would save lots of time, would it not? Fortunately, this is already possible due to Electronic System Level (ESL) design, which is compilation of techniques that allow to design, simulate and partially verify a digital chip, all within any normal laptop or desktop computer. Moreover, some ESL tools such as the one investigated in this project, allow you to even simulate a program code written specifically for this hardware; this is known as virtual hardware software co-simulation.

The reliability of simulation must however be considered when compared to a traditional two-step methodology or FPGA board usage to verify a full system. This is because a virtual hardware simulation can have several degrees of accuracy, depending on the specificity of component models that make up the virtual prototype of the digital system. Therefore, in order to use co-simulation techniques with a high degree of confidence for verification, the highest accuracy degree should be employed if possible to guarantee that what is being simulated will match the reality of a silicon implementation. The clock cycle-accurate level is one of the highest accuracy system simulation methods available, and it consists of representing the digital states of all hardware components such as signals and registers, in a cycle-by-cycle manner.

By using the ARM SoC Designer ESL tool, we have co-designed and co-simulated several microcontrollers on a detailed, cycle-accurate level and confirmed its behaviour by comparing it to a physical reference target development board. Finally, a more complex virtual prototype of a cellular IoT system was also simulated, including a micro-controller running a a real-time operating system (RTOS), hardware accelerators and serial data interfacing. Parts of this virtual prototype where compared to an FPGA board to evaluate the pros and cons of incorporating virtual system simulation into the development cycle and to what extent can ESL methods substitute traditional verification techniques. The ease of interchanging hardware, simplicity of development, simulation speed and the level of debug capabilities available when developing in a virtual environment are some of the aspects of ARM SoC Designer discussed in this thesis. A more in depth description of the methodology and results can be found in the report titled "Virtual Cycle-accurate Hardware and Software Co-simulation Platform for Cellular IoT". (Less)
Please use this url to cite or link to this publication:
author
Tovar Rascon, Marcel LU and Elfborg, Patrik LU
supervisor
organization
alternative title
Virtuell och klockcykelexakt samsimuleringsplatform för hård- och mjukvara till mobilt IoT
course
EITM01 20162
year
type
H2 - Master's Degree (Two Years)
subject
keywords
co-simulation, cycle-accurate, ESL, HLS, SystemC, SoC Designer, IoT, SoC, FPGA, embedded
report number
LU/LTH-EIT 2017-602
language
English
id
8926938
date added to LUP
2017-10-17 08:47:22
date last changed
2017-10-17 08:47:22
@misc{8926938,
  abstract     = {Modern embedded development flows often depend on FPGA board usage for pre-ASIC system verification. The purpose of this project is to instead explore the usage of Electronic System Level (ESL) hardware-software co-simulation through the usage of ARM SoC Designer tool to create a virtual prototype of a cellular IoT modem and thereafter compare the benefits of including such a methodology into the early development cycle. The virtual system is completely developed and executed on a host computer, without the requirement of additional hardware. The virtual prototype hardware is based on C++ ARM verified cycle-accurate models generated from RTL hardware descriptions, High-level synthesis (HLS) pre-synthesis SystemC HW accelerator models and behavioural models which implement the ARM Cycle-accurate Simulation Interface (CASI). The micro-controller of the virtual system which is based on an ARM Cortex-M processor, is capable of executing instructions from a memory module.

This report documents the virtual prototype implementation and compares both the software performance and cycle-accuracy of various virtual micro-controller configurations to a commercial reference development board. By altering factors such as memory latencies and bus interconnect subsystem arbitration in co-simulations, the software cycle-count performance of the development board was shown possible to reproduce within a 5% error margin, at the cost of approximately 266 times slower execution speed. Furthermore, the validity of two HLS pre-synthesis hardware models is investigated and proven to be functionally accurate within three clock cycles of individual block latency compared to post-synthesis FPGA synthesized implementations.

The final virtual prototype system consisted of the micro-controller and two cellular IoT hardware accelerators. The system runs a FreeRTOS 9.0.0 port, executing a multi-threaded program at an average clock cycle simulation frequency of 10.6 kHz.},
  author       = {Tovar Rascon, Marcel and Elfborg, Patrik},
  keyword      = {co-simulation,cycle-accurate,ESL,HLS,SystemC,SoC Designer,IoT,SoC,FPGA,embedded},
  language     = {eng},
  note         = {Student Paper},
  title        = {Virtual Cycle-accurate Hardware and Software Co-simulation Platform for Cellular IoT},
  year         = {2017},
}