 Hello all. I hope you all are safe and healthy. I am Ravi and I am working at Samsung Semiconductor in India currently. Today I will present a session on Automotive Ethernet, which is emerging in today's automotive world and collector vehicles. Let me go through the overall agenda of the session. First, we will go through the overview of automotive systems and its evolution. Then we will see what are the different technologies which are used in today's automotive networks. We will see the different network protocols like lane, cam, flexory, most and so on. Then we will discuss about the future requirements in the networking domain of automotive network as complexity is increasing continuously over the time. Then we will see the motivation which leads us towards the need of automotive Ethernet in spite of there are so many protocols which are existing over the last decades. Later, we will discuss why conventional Ethernet cannot be directly used in the automotive industry. In the next slide, we will see the history and the evolution of automotive Ethernet from where it started and what are the different technologies it evolved over the time. Then we will see the protocol format of the automotive Ethernet. In that we will see the data lake layer and IP layer protocols. Then we will discuss the different technologies which is used in the automotive Ethernet like autosar, open, power over Ethernet, energy efficient Ethernet and so on. Later on, we will go through the support of automotive Ethernet inside the layers. In that, we will see how the time sensitive network is supported in the layers. Then we will see the different checked physical layer in that we will see the physical layer overview of automotive Ethernet. At last, as a conclusion, we will discuss the overall benefits of the automotive Ethernet and what are the different challenges which automotive industry is currently facing. Let's start with the overview first. The automotive industry has very long history which was initially started with the steam engines. Then later internal combustion engines came then nowadays we are seeing a hybrid and electric versions. The normal car contains too many moving parts so it contains too many actuators and sensors. To connect and control them, controller is also present. During initial this, all these components are connected by normal wires. Now think about the requirements of wires and its complexity. If we use the normal wires to connect them. That's how the various vendors and groups has developed the different protocols to fulfill those requirements of automotive data communication. Now if you have seen the car bonnet, then you must have seen the wires connected at the different wires. This wiring is called nothing but the wire harness and it is the third highest cost component of the car because of its complexity. In fact companies try to skip the wire harness under the standard vehicle warranty in order to reduce the company burden. We saw how complex is automotive network. Now let's look into the recent trend in automotive network. As we know the automotive vehicle and ECU or engine control unit is present. Which is basically controls the vehicle function related to engines, import and maintenance systems and newly trending ADAS system. As you can see in the image that the different protocol are used in typical automotive vehicle depending on the usage and the bandwidth requirement. Now let's look over the different protocol one by one quickly. The first one is a link which is local interconnect network which is announced in late 1990. It basically provides a very low cost network in a vehicle. It is a single master bus and provides the speed up to 20 kbps. Since it is a single master it is mostly used to connect the sensors and the controller. Then next comes can which is a controller area network and it is known as it is known for the its robustness and safety. It basically developed by the Bosch in 1986 and it is a multi master protocol and it provides a data rate up to 1 mbps. Since it is a multi master it is used to connect different controllers and controller to sensors. Depending upon the requirement there are some variations created over the time like TT can, can FB. The next guy is a flex rate which is an alternative of high speed can and it provides more reliability and more speed compared to the can. But it is also more expensive than the can. It provides a speed up to 10 mbps and it is a multi master protocol. It can have two independent data channels in order to provide the $1. The next one is a most which is from media oriented system transport. As the name suggests it is a high speed multimedia protocol which is used to transfer audio, video, voice and data transmission. It is mostly used in the infotainment system. It provides the bandwidth up to 25 mbps and it can go further up to 150 mbps. At last comes Ethernet which is a packet bridge protocol and it can support the speed up to 10 mbps. There are multiple standards available at the different layers to support different functionalities. We will focus on automatic Ethernet in upcoming slides. Apart from this protocol there are other protocols exist in the market but it is not that much popular. As we seen in the previous slide that there are multiple protocols exist for a different type of use case but the development and the need of advancement in automotive industry has not stopped yet. Nowadays everyone wants a car with advanced feature like ADAS which includes adaptive cruise control, around wing monitor, lane detection, automatic braking and so on. Also we want all these features which are easily controllable from the single dashboard and without losing the reliability and the safety of the car. So these future requirements create another problem like increasing the number of wires and its complexity. It also creates a problem to integrate these new features into the single software and providing access to the dashboard without losing the reliability and safety of the vehicle. The requirements are basically never endless in the world. Then why we need automotive Ethernet? What is the motivation behind it? Well there are some of the things which basically involve the concept of automotive Ethernet. As we know in automotive industry that sharing all the data requires a higher bandwidth. Now you can imagine a case of lane detection where single sensor requires a bandwidth up to 70 mbps. Now here the Ethernet is a famous protocol which is providing a speed up in gbps. Automotive Ethernet is a very safe and cost-worthy network compared to the other protocol. It provides significantly higher data rates so we can combine multiple CAN bus into a single Ethernet connection. So wiring and installation costs are greatly reduced. Automotive Ethernet provides different features like quality of service and sensitive network. So real-time communication and transmission of low priority data at the same time is possible. It has advanced security features which can protect highly connected vehicle from hackers and viruses. It can be easily integrated with the existing TCP-IP protocol. Automotive Ethernet has a very low silicon cost and it requires very less space for wires and the connectors. Then question comes into the mind that why we cannot use normal Ethernet in the automotive industry. So there are a number of reasons behind that. The very first thing is that a normal Ethernet requires at least 4 wires which is not acceptable here because our target is to reduce the amount of wiring. The second thing is that there is a high radio frequency interference and EMI is present inside the car which normal Ethernet cannot be able to tolerate. As we know that a normal Ethernet uses CSMA CD mechanism to detect the carrier and transmit the data so it cannot provide guarantee latency in terms of microsecond. There is a requirement of bandwidth control for different streaming so the normal Ethernet cannot be fulfilled. The precise time synchronization is not exist at the protocol level of normal Ethernet. The temperature of the car can go up to 125 degrees centigrade and the acceleration of the car can go up to 4G which conventional Ethernet cannot be able to handle. The normal Ethernet does not complain with the automatic safety or ACA level certification. There is a need of low power standby mode during the engine of condition and instant wake-up is also required to unlock the vehicle quickly. So this type of requirement the normal Ethernet cannot be able to fulfill. Now let's see the history and the evolution of automatic Ethernet and how it is evolved for the different generation. Now automatic Ethernet is a subcategory of Ethernet which is specified in the IEEE 8.2.3 specification. It basically operates over a single pair of twisted wires and it is specifically designed for the low-radiated emissions and immunity requirement of the automotive industry. The operating distance is a much shorter than the normal Ethernet. As you can see in the picture that automatic Ethernet has almost crossed 3 generations from the year 2010 to till now. In the 2010 it started with the use of Ethernet in cars for the diagnostic and firmware update. It is using 100 base tier standard, though it does not meet the automatic requirement but since this interface is only used for the diagnostic and service purpose an exception was added for its usage in the car. Then in 2015 the advanced driver assistance and infotainment systems came into the picture. In that time Ethernet is used only for point-to-point links. It is not used as a shared medium for different purposes. In that time frame the different routes like border range and audio-video bridging has been released the different standard for the interface of automotive Ethernet. In later stage it has been extended to support more bandwidth up to 1G DBS and Ventus has now started using the Ethernet in a full phase mode. Here the second generation of AVV standards has been started evolving. The new advancement not only helps to reduce the cost and weight but it also makes easier for the different system to cooperate in the car. Now let's see the frame format. It is almost same as commercial Ethernet. The very first is Ethernet. So as we can see the first 7 byte which is known as preamble bits that contains the pattern of alternating 0s and 1s. It indicates at station side that frame is going to be started and it also enables the sender and receiver to establish a bit synchronization. The next one byte is a SFD for a start-up frame diameter bit which is always set to some fixed sequence of 0s and 1s. The last two bits are always 1 which is indicating that end of SFD field and marks the beginning of the current frame. The next comes destination address which is a 6 byte field and contains MAC address of the destination for which the data is going to be sent. And after comes the source address which is also a 6 byte field and it contains the MAC address of the source which is the sending the data. The next two bits are reserved for the length which specifies the length of the data field. Here this field is required because the Ethernet is using variable size frame. Then after data fields came into the picture which size can go from 0 byte to 150 byte and it contains the actual payload of the frame. Now as per the specification the minimum payload size of the frame must be 46 bytes so a varying field is required according to the size of data. At the last the frame check sequence or FCS is set which is a 4 byte field and it contains the CRC code of error detection. Now next is the IP layer which is also the same as conventional Ethernet. At the IP layer the IP packet is made up of two parts the IP header and the IP body. From a packet filtering point of view the IP header contains very interesting piece of information. The IP packet contains various field like source address, batch address and to live type of protocol offset of the packet, IP version, flag and so on. If a packet is too large to cross the given the nature the IP packet is divided into smaller smaller packets which is called as a fragments. The fragmenting of the IP packet does not change its structure at the IP layer but the body of each IP packet contains only small piece of information. Now let's dig into the some different technologies of Automotive Ethernet. First we will go through AutoSAR which is known as Automotive Open System Architecture. AutoSAR is formed in column 3 by major Automotives. The main objective of AutoSAR is to create and establish an open and standardised software architecture which can be used in Automotive Electronics. The AutoSAR basically provides the specification of basic software module which defines application interface and builds a common development methodology based on the standard data exchange format. Its software architecture can be used in vehicle of different manufacturers and electronic components of the different suppliers so which in turn reduce the cost of research and the development. Based on this principle AutoSAR is mainly focusing on upcoming technologies. Now AutoSAR is a basic three-layer software. The first layer is a basic software which is a standardised software module and the other service which can be used as the functional part of the upper layer software. The second layer is a runtime environment or RTE. It is a middleware software which is extracted from the network topology of ECU information extent layers which is present between the application software component and between the basic software and the application. The last layer is the application layer which contains application software component which is interacting with the runtime environment. Now next technology is the open which is a short form of openware Ethernet. It is basically an alliance which is a non-profit and open industry alliance. It contains automotive industry and technology providers. It encourages the wide scale adoption of Ethernet based network which can be used as a standard in automotive networking applications. It basically enables the development and the existing IEEE hundreds based T1, thousand based T1 and thousand based RH physical layer specification. Its goal is to complete the ecosystem further with the requirement and test specification for hardness which is used and additional functionalities. It encourages and supports the development of new physical layer solutions in a standard setting organization. It continuously identifies and address the gaps related to the implementation of Ethernet based communication in automotive domain. Now next technology comes is a POE or power over Ethernet. The POE is a basic technique to provide the DC power over the existing Ethernet cable. So the requirement of separate power line and outlet is removed. It was basically standardized as the IEEE 8.2.3 AF in program 3 and specified that the power can be either provided by the spare wires or the data wires. The standard basically also includes a mechanism to protect device which does not support the POE mechanism. A standard 25 kilowatt resistor is added between the power pairs at the power device size and the power source provides the power only if something similar to 25K resistance is detected. Carrying both power and the data across a single cable not only reduce the cabling needs but it also improves the safety and simplifies the installation which in turn helps the saving time and reducing the cost. At the later stage the power supported by the POE is increased to support a wider variety of end device. So there are several additions and improvements made over the original specification. Now we will see about the POE which is a short form of energy efficient Ethernet. POE is a basically set of enhancement over the standard Ethernet physical layer that reduce the power consumption during the period of low power, low data activity. The main intention of POE is to reduce the power consumption by almost 50% or more while retaining the full compatibility with the existing equipment. In a fast Ethernet like Gigabit Ethernet or 10 Gigabit links the constant and significant energies used at the physical layer as the transmitters are always active whether the data is being sent or not. This range of power consumption can go around one-on-one which is a significantly higher. As you can see in the image that when there is no data to be sent the software puts the physical link into the low power mode by issuing a low power IDLE request to the Ethernet controller. The file layer then sends the LPS symbols to the transmission link for a specific amount of time and then it disables the transmitter. Now to maintain the link activity a refresh signal is going to be sent at a specific form of intervals. Again, when there is a data to transmit the normal IDLE signal is going to be sent for a specific amount of time. So that indicates at the receiver side that the data transmission is going to be start. Here the data link is always considered as active because the circuit at the receiver side remains active even when the transmission data line is in slip mode. Now what is a low power IDLE? The main concept of low power IDLE is to transmit the data as fast as possible and immediately return back to the low power IDLE mode. It basically saves the energy by switching between active and low power IDLE and the power can be reduced by turning on use circuit during the low power mode. Now let's see the time synchronization which is one of the part of time-sensitive networking. The TSN is basically a group of IDLEs which is formed in the 2010. Before TSN, there is an audio-video bridging group is present and later on it is renamed to the TSN for making it for wire scope. In the TSN, there is a very well-known IEEE 802.1 AS protocol which defines the synchronization of timing among the nodes connected in the network. In this protocol, there is a based master clock algorithm which decides the master clock and the master node. Master node distributes the clock information to all IEEE supported nodes. This protocol is able to support different network technologies such as IEEE 802.3 EPON and IEEE 802.11 wireless LAN. But the transfer and the synchronization message depends on the network media due to the difference time and mechanism. We will see in the detail about this in upcoming slides. Now, time synchronization is used as a time-critical application like ADAS which includes adaptive cruise control, emergency brake assist and so on. There is another important protocol which is nothing but diagnostic over IP or simply DPO IP, which is widely used to analyze the data from onboard components and to update the firmware of ECUs and TCUs. The protocol is based on ISO 1340 standard which specifies the requirement of diagnostic communication between external gate, test equipment and vehicle electronic components. As an example, while servicing your car, you must have seen the technician with the OBD connector which has a very small display and they are basically attaching it to your car to rectify your car issues. Those OBD connectors are nothing but the example of DOIP. Sometimes the technician also connects the laptop for the firmware updates. The DOIP allows faster data rate at lower cost compared to conventional Canvas diagnostics. As we discussed that one of the advantages of automotive Ethernet is it can be easily integrated with the TCP-IP protocol and it provides the open source software support. So, let's see the support of automotive Ethernet in the latest. I will discuss the time-sensitive networking in upcoming slides as time-awareness is one of the key differences between conventional Ethernet and automotive Ethernet. Now, let's see the time-sensitive networking. The TSN is a basic list set of standards which is developed by the IEEE 802.1 group. It is called as IEEE time-sensitive networking task group. Before TSN, the ABB group was present, they defined some standard to stream or transfer the audio and video over the Ethernet to meet the bandwidth and latency requirement. The TSN was formed in 2012 by adding some new standards and upgrading existing of the ABB group. As you can see in the image, the TSN focus on four different parameters. The first one is the time, which means that all devices connected to the network should have same time information. Next comes synchronization, which means that all nodes in the network have sub-microsecond or nanosecond level of accuracy all the times. It ensures precise controlling of the node. Now, you can imagine a case in automatic network like if an engine controller decides to apply, brake on four four wheels and if there is a small delay between any of the brakes then how the serious scenario it can be. Now, next is a traffic shaping, which means that time critical messages on the network should be reached to the target destination irrespective of interference from other traffic. As an example, when you apply the emergency brake then it should be get applied immediately. The last is a boundary latency, which means that the messages traveling in the network should reach the destination within a specific time and the latency should be guaranteed. As we have studied in our academic data, there are basically seven layers of OSI system in the reference implementation, but in practical, only four or five layers are exist. Now, the TSM sits on layer two, which is a data link layer. The TSM requires support at hardware and software both for the proper functionality. Rest of the upper layer protocol like IP layer and TCP-UDP layer mostly remain unchanged. Now, being TSM at data link layer we can see the delay and the interference very less, which helps to achieve timing and latency requirement of the specification. Now, as we can see in the previous slide, that TSM is the set of standards and extension of existing AVB group. So here, you can see the blue mark boxes, which is the part of AVB group and the gray mark boxes, which indicates the extension of newly developed TSM standard. The 802.1 ES is known as the precision time protocol for simply PDB and it is first standardized in 2002. The aim of PDB is to provide a mechanism of syncing the absolute time over the standard requirement. The basic PDB protocol has gone through several division, but there are almost two nominal version. One is the I3 Poly1500 series, which is commonly named as PDB. And the second one is I3 Poly802.1 AS series, which is known as GPTB or generalized PDB. The I3 Poly802.1 AS has reduced the number of supported options in order to improve the better performance. As you can see in the image, that I3 Poly802.1 AS is the latest standard and it is the part of TSM. Its goal is to tighten up the performance requirement to the next level to accommodate the emerging TSM standard. For resource management like streaming, the I3 Poly802.1 QAT was standardized in 2010, which is known as a string reservation protocol or simply SRB. It defines the concept of string layer at layer 2 OSI middle. There is a listener and broker kind of concept in the streaming. Later on as an enhancement of 802.1 QAT and I3 Poly802.1 QCC was formed in 2018. The basic aim was to provide better quality of service over the existing one. For the transfer string and control, there are basically two protocols, I3 Poly1722 and 111221. I3 Poly1722 is the layer 2 audio video transfer protocol which was formed in 2011 and it defines the details of transmitting specific strings over the other AV formats. It sets the presentation time for each AV stream and manages the latency. On top of that, I3 Poly17172.1 allows AVB discovery, generation connection management and control of the devices. It is basically formed in 2013 from existing 1722 protocol. Both I3 Poly1722 protocol bypasses the middle layer standards of networking style. So there are no enhancement made in the TSN. For traffic security, I3 Poly802.1 QAB provides the forwarding and the buffering of time-sensitive packets. It was standardized in 2009 and provides a guarantee and loss-sensitive real-time audio video data transmission. The credit-based shaper or CBS which is one of the earliest traffic shaper which was standardized in the AVB. After that, the different traffic shaper was standardized in the TSN like I3 Poly802.1 QCH which introduced double buffering concept and it allows bridges to synchronize transmission in a cyclic manner. Then next came I3 Poly802.1 QBB which separates the communication on the network into the fixed land and repeating time cycles. We will see in upcoming slides in this detail. For asynchronous network traffic, I3 Poly802.1 QCR is a form and it uses local clock in each bridge to manage the traffic. The standards I3 Poly802.1 QBU and I3 Poly802.1 BR is used for preemptible traffic. In that preemptible frame is interrupted to pass the express frames. There are other protocols like 802.1 CB 802.1 QCA and 802.1 QCA which defines the method for fault tolerance and it is one of the important aspects of the TSN. In next slide, we will see the popular protocol like VTP and traffic schedule in detail. As we have seen in the previous slide that the time synchronization is a standardized I3 Poly802.1 AS protocol. This protocol is implemented in the Linux as a Linux VTP. As we have seen in the previous slide that VTP protocol use the based master clock algorithm to determine the master clock in the E1 network. And then after the master node which is determined by the master clock which distributes the clock information to all other nodes. This sharing of the clocking is done at the data link layer and the hardware. So the clock accuracy is achieved in terms of nano-signals. There is another similar protocol which is known as a network type protocol or NDB. And it runs basically on the network layer but it provides less accuracy. We do the addition of the software layers. Now in the Linux ecosystem Linux VTP is the most popular implementation of VTP. It supports several profiles including GPTP and heavy and new automotive profiles. Linux VTP provides some utility tools to carry out the time synchronization. The first one is the VTP4L which is nothing but a demo that synchronize VTP hardware clock from the NIC. The next came PHT2Cs which is a also demo that synchronize the PHC and the system clock. The last is a VMC which is a utility to configure the VTP4L in the run time. Though VTP is a user leverage step the kernel provides some component to access the VTP hardware as shown in the figure. It supports the hardware and software time stamping via the SO time stamping socket. And it deals directly with the hardware using the Linux VTP hardware clock or simply VHC subsystem. One thing to be noted here is that all nodes must support the VTP mechanism in order to synchronize the clock effective in the network. Now let's see how traffic scheduling is handled in the Linux. In the TSN, the traffic scheduling is standardized as the I3.8.2.1 UBV standard. In the Linux, there is a traffic control system or PC utility which is responsible for handling of network traffic scheduling. It uses transmission algorithm specified by FQTSS and it is supported by DC queuing, disciplines or simply QDisk. The term FQTSS is used to describe setup tools which are used to forward and queue the time-sensitive streams. A QDisk is a scheduler which controls when to send a packet to the stream and when to pass the packet to a player. And this is basically defined by the different parameter which is configured by the DC utility. There are basically three QDisk which are supported in the Linux TSN. The first one is a CBS QDisk which determines the credit rates mechanism and it is introduced by the I3.8.2.1 QAB amendment. It basically seduces the transmission according to how the bandwidth for the credit which is reserved for a given outbound queue. Another QDisk is a time-over priority shaper for simply a pre-op. It implements the I3.8.2.1 QBB standard. It works on a basic concept of time division multiple access where the channel is divided into small small time slot and the different time slot are assigned and configured based on the priority. The third one is the earliest EX time first for simply ETF QDisk. It basically implements the frame to be transmitted at a specific time frame. In the Linux this hardware feature is enabled to the SOPX time socket and the ETF QDisk. Here the SOPX time socket option allows application to configure the transmission time for each frame while the ETF QDisk which ensures that the frames coming from the multiple socket are sent to the hardware ordered by the transmission time. So far we discussed all assuming that the network has only N device but that's not true, right? It might possible that network consist one or multiple switches and switch terminals. So to support this type of scenarios NFS provide basically two frameworks one is a DSA or Distemper Switch Algorithm and second one is a Switch Devo. The DSA is initially introduced by the Marvel in 2008 to support their own switches. Later on other vendors have also added DSA in their own IPs and contributed in the Linux code and that's how the DSA has been grown. Ethernet switch is basically having a multiple front panel ports and the multiple CPU or management ports. The DSA relies on the management port which is connected directly to the Ethernet controller. Here the Distributor parts comes from the possibility that to have multiple switches connected together through the dedicated ports. The main idea behind the DSA is to reuse the available internal representation and tools to describe and configure the switches as a common logical component. In contrast to DSA the Switch Devo is a more recent approach which focusing on outsourcing the maximum possible work to the hardware. This way it enables the replacement of propriety SDKs and software with a standard open Linux interface. This approach is effective for both server side on smart NICs as well as routers and switches. This allows end user to remove the dependency of vendor specific APIs. We cannot consider Switch Devo as a traditional Linux device model. Now let's look for the overall summary of TSM. So far there are many IEEE standard which are available under TSM but not all of them are implemented in the Linux currently. Only few of the protocol like time synchronization, traffic scheduling and framing are supported under the Linux either fully or partially. The main reason is that it was already a part of AVB and it is a very old. The different vendors have also sent the pages for various TSM implementation but those are either under review or draw. Though the majority of the protocol are implemented or supported as a Linux functional natives. One of the example is AVNV which is a group of the member companies and it is working together for the TSM in each solutions. Let's see the quick overview of physical layer of automotive The initial standard was officially released in 2011 by Borda Rich Group which is formed by the multiple companies and mainly promoted by the dotcom corporation. Later in 2015 it was standardized by the IEEE under IEEE 802.1 BP which is known as 100 base T1. As you can see in the image that it uses two wires in a prister pan form instead of four wires which we are using in the conventional Ethernet. Here in T1 the one denotes that it is a single prister pair of wire. Later on a new standard was added for higher bandwidth as IEEE 802.1 BW and it supports the 1000 Mbps build and which is known as the Gigabit Ethernet. Here you can see the basic characteristics of 100 base T1 and 1000 base T1. Both are supporting a full base mode and multi-level VAM 3 coin. Means there can be 3 bits for the symbol. Both support the bandwidth up to 600 MHz. The maximum length of the cable can go up to 15 meters while Gigabit Ethernet can support up to 40 meters of cable with an optical fiber. Here the 100 base T1 supports the additional features like energy efficient Ethernet. Let's conclude the session. We have seen the automotive vehicle from the last 3 to 4 decades and it has become our part of the life which we cannot imagine the world without it. From initial days to now, automotive vehicles have become complex and more user friendly but as a side effect the amount of cabling has been increased significantly over the time. So if we can reduce and simplify the cabling at the multiple level like fuel consumption is reduced due to the reduction of the weight. The issues while preparing or servicing is reduced. The cost of manufacturing of vehicle is also reduced. Now and less time is required to assemble the vehicle. Here automotive Ethernet fulfills all the criteria and there are active contributions from the different vendors and the organization at software level to make it more robust. But there are also some challenges which industry is facing like timing and cost are high for the development of testing of ECU because there are dependencies of vendors and multiple certifications are involved in the development cycle. As automotive Ethernet directly integrates with the common EC by step, the security flow is added by report. In order to make it more efficient, I recommend that interference, robust cabling is needed to save lots of data. For redundancy if we increase the error correction bits then bandwidth is reduced and if we increase the bandwidth then error correction bit is reduced. So yeah, that's it from my side and I am happy to address your question and thank you for attending the session.