Overview of Embedded Systems Fundamentals

[Music] okay good morning and welcome to you all to this course on embedded systems we will try to understand in today's lecture which is the first lecture how embedded systems differ or how embarrassed system poses some additional constraints over the other general-purpose systems that we have encountered in our learning process till now now embedded system what is an embedded system if we ask this question what is an embedded system the possible answer is an embedded system is a system where a microcontroller based or microprocessor based programmable system is embedded in a larger system so you will have a larger system which can be this is the larger system in which there is a microcontroller based or programmable system embedded always it may not be programmable by the way it may be also hardware programmed okay so there is a some processor embedded inside this and which interacts with this larger system at different levels so this system actually takes inputs from the world and by the world I mean this environment with a world for this embedded system so it takes inputs from this world processes the information tense it back to the world it can come from different points of this larger system and the after computation the responses can go to again different points of this system the points which are affected because you see the larger system can have many things everything may not be controlled by the embedded system everything may not be sensed by the embedded system so the parts which are sensed acted upon and actuated upon are known as often known as the sphere of control so might be there is a sphere of control around this embedded this embedded system within this larger system that is being affected by this embedded system all right now typical examples of embedded systems now you can think of embedded systems are everywhere everywhere you can say if you take your mobile phone it has got many communication systems but also the other processors even earlier if you take the printer inkjet printers or color printers they're also the controllers are there even if I go even earlier there are some disc systems where intelligent controls are embedded okay printers have already said so there are there have systems have been embedded inside larger systems for longer time nowadays we find intelligent washing machines where depending on the volume of the clothes the extent of dirt in the clothes the gyration the speed at which the will be centrifuged or how much detergents will be required all are controlled automatically how is it controlled automatically using some embedded systems again if we go to more advanced things now most of the modern cars are coming with not one embedded system but multiple embedded systems okay which some are taking care of controlling the brakes some are taking care of the cruise control some are taking care of the locking system so there are number of embedded systems impregnated inside the larger system that is the car if we look at the airplanes most of the airplanes now are said to be not so much dependent on the pilots right so there is a set of embedded controllers inside that which can negotiate with the environment okay and those are also embedded systems in warfare when we find the when we want to have some rocket launchers they're also embedded systems are there which are taking signals from here so the list gradually becomes endless nowadays now if we look at now this is also known as a sometimes a variant of such embedded systems are known as cyber physical systems this term is becoming popular day by day whether cyber part means the computer interacting with a physical environment so the physical environment can be a road on which a car is moving right the road will have different frictions at different points that is a physical reality and now I have got a car which has got a controller which will keep the speed fixed at a particular set level now from your school level knowledge of physics you know that in order to keep the speed constant when the friction increases you have to apply more force and the reverse when the friction goes down in order to keep the same speed now that the physical reality is the road the physical reality is a condition of the tires of the car and the controller is within the car this is one sort of thing once out example of cyber-physical system this can be further extended nowadays to the term which many of you have heard which is internet of things iot as it is told IOT is nothing but a set of systems spread around maybe distant each of them have got their embedded controllers embedded sensors everything and they are communicating over the Internet and one system can communicate and actuate another one at a distance therefore everything gets connected through the communication of a set of embedded systems okay now coming back to embedded systems proper if we think of an embedded system it will consist off say a microcontroller based system or processing system let me call it the processing system which usually our microcontroller or microprocessor based if they are programmable and on the other side we have got the sensors which interacts with the world and provides the data to the processing unit and there are actuators okay actuators now these actuators are actually affecting the physical world or the larger system let me call it the other system whatever that other system might be alright and the sensor is sensing some variables from this other system and feeding it to the processing unit and depending on whatever actions are to be taken there the actuator is actuated it now in between I have missed out certain things intentionally that is it can the sensors are sensing the data which are coming from the other system which is the physical world which is mostly analog alright mostly analog right if it were an analog system therefore the there can be two things the sensor can directly sense the data in the digital form otherwise if it does not do that then here we need a layer of interface which is the analog to digital converter because this processing system is a digital machine similarly we have to give the data to the physical environment mostly analog may or may not be but if it be an analog system then we need here addie to a converter digital to analog converter right now often the actuators can be directed digital and can provide the control that we'll see some examples later alright so this is the overall environment so in order to learn embedded systems we have to learn some we have to have some knowledge about the sensors what sensors how do you specify sensors may be many of you know how analog-to-digital converters or digital-to-analog converters are working we will try to brush up on that he also in and most important part is this processing unit where we will see that different types of processing can be involved now you one can ask the question that in a typical architecture of a computer system we also have got the input device the CPU and the memory the processing device pricing block and the output device so where does it differ from a general-purpose system so how does it differ from the general-purpose system so general-purpose system versus embedded systems usually the embedded systems are one embedded system is for one purpose so single purpose okay or single application that is point number one the other point is the embedded systems is severely usually what happens here what do you mean by this a single application program is repeatedly run whenever for a particular say for example let me give the example of an answering machine telephone answering machine it is going to fix task that when the call will come after five or seven rings if the receiver is not picked up see all these things that we sensed that the receiver has not been picked up then the answering machine goes on after the delay gives of a flute plays a message and gives a bit after recording for a particular delayed cuts it off all those things now this is a fixed program it has been written once and have been put in an embedded processor which has been put in the answering machine or the telephone system so that program is repeatedly working the other option is other distinguishing feature is it is very tightly constrained compared to the general-purpose system general-purpose system is not single perp single purpose it will be multiple purpose it can be used for different purpose now is tightly constrained in terms of what it must be low cost the cost cannot be exhibited exorbitant like a general-purpose server right it should be low power mostly we would like to run this on battery it should be portable ok it should be sometimes it must be real time in general in general it should be fast it should be fast all the time but sometimes it should be hard real-time I will come to that what it means okay so these are some of the constraints that are specific to embedded systems but not so much for general-purpose systems on the other hand general-purpose systems are more powerful in the sense that it can do many more things it is not single purpose it is not constrained by these things it consumes more power usually runs from the direct electric supply with some battery backups and all those so that is the differentiation between general-purpose systems and embedded systems most important thing is it is predefined what I am going to run on an embedded system is predefined application in some cases you will find that multiple applications are running on an embedded system but those are also predefined three or four it cannot be just like a general-purpose system that you can go on adding to this right I will differentiate between another thing that since right now I have said real-time systems let me also differentiate between RTS and embedded systems what our real-time systems real-time systems our systems it can be of two types hard hard real-time systems and soft real-time systems harder TS and software tears now hard drive real-time systems have got a stringent time constraint the time constraint that is given is very stringent now this stringent time constraint must be adhered to must be respected if this time constraint is violated then violations will lead to catastrophic you have to land within a plane has to land within maybe one minute all right now if it exceeds one minute then it can crash take another example say a nuclear power plant or some power plant some thermal plant where you find that the temperature has gone very high beyond the safety limit and you have to start the coolant and put the power put the shut off the plant within some specific time and suppose your system that is controlling it is sluggish enough that it does not respect that time it does the thing but does it much later by that time the situation has blown off right so that is not allowed in hard real-time systems why do you call it real time systems because this time constraints are specified in terms of real time real time means what one second to second by this clock this is a real time clock time is ticking on the other and what is the other sort of time that we discuss about we discuss about so many cycles suppose I say that the light must be switched off after 50 cycles vasavi I say that the light must be switched off within one second or one minute a one second what is the difference if the cycle clock is 50 Hertz then they are equivalent because i have said after 50 right so it will be fine but if the clock is slow then 50 cycles will come after two minutes right so that is the basic difference between real time systems and normal I mean that typical computer systems that we talk of on the other hand hard real-time soft real-time systems if we violate the time constraint then we the quality of service gets reduced quality of service reduced but no catastrophe pans no catastrophic happens ok now embedded systems can be real time systems or may not be real time systems all right it may be a real time system or may not be a real time system again on the other hand similarly real time systems may be embedded usually real time systems are embedded but real time systems can also be say a server based system okay so these two are not equivalent that all embedded systems should be real time systems that should be kept in mind next we'll come to a slide that will give you an example of an embedded system here so this is an example of an embedded system which is a digital camera it has been taken from the book embedded system design a unified hardware software introduction by Frank vahid and give urges ok it is a wily publication so here you see typical digital camera what do we have inside now first of all it is a single function it's single function is only one function that is being done that is photograph is taken and stored right it is tightly constrained how it is low cost otherwise people will not bite low power it is small portable must be portable and should be fast now we do not call it real time system if the picture is captured a little slow no catastrophic happens what it is reactive and real time i will explain what is a reactive system and real-time system but it is a reactive system and real time only to a small extent because if you take the image it is a soft real-time or hard real-time if soft real-time obviously because if it is very slow then probably will not buy this ok but no catastrophic will happen now what is there you can see the lens which is capturing the world through a CCD charge-coupled device there is an A to D converter which is converting it there is a charge-coupled device preprocessor then here you see one thing that I in the slide that I was showing about the embedded system architecture all through you can see there is a communication channel going on all through this that I did not explicitly mention but communication is another very important part of this entire embedded system ecosystem so here you see that through the bus they are communicating all right this is Sydney Prosser data is going everywhere and there is a microcontroller there is a JPEG codec which is doing the conversion and compression and also there is a dma controller memory controller bus interface eart is a serial parallel to serial converter okay and there is an LCD control so there are so many components inside an embedded system besides the process now you see how many processes do you see here you can see a microcontroller as a processor here you can see your CCD p processor pixel coprocessor here is and the jpeg processor multiplier you can say that this is not a price i j-just a multiplier but that is also doing some privacy so there are so many processes communicating now the key point is that some of these can be hardware can be implemented in hardware some of them can be implemented in software alright we will see this part next so what are the design challenges before i come to the design challenges let me make another system level classification one is reactive systems versus transformational systems reactive systems are those which react to an event that occurs for example there is a system like a camera it does not of itself work whenever you create an event you there is an input and in that input you put a pulse like suppose you are pressing the shutter then only it starts working similarly it could be that this is continuously it is keeping watch it is active another system but whenever there is an alarm situation then only it starts working and immediately it sends the response and usually such reactive systems you in many cases such reactive systems react to an event which is an argent event and therefore in many cases they are real time systems on the other hand so these are working on external events on the other side there can be transformational system for example image processing systems you get a set of images you get an image and you process it in image processing you do a lot of things you do noise removal you do segmentation right and you do go on doing all these things different activities that you do right now that you are taking a data some data is coming in and that data is being transformed through stages not in the event of any not in the event of some explicit event asking for service okay here if we go to this diagram you can find that there will be some reactive components also here and some will be transformational components so often the system is a mix of these okay now embedded systems have got a number of design challenges major design challenges the cost obviously cost is a very important aspect now the cost has got two components one is the energy costs energy cost means non-recurring engineering costs okay you have to we will have to incur some cost on that so here are what are the challenges the challenges are mostly mostly the challenges are optimizing on different aspects okay one is the cost I mean how much time how much how much money I will spend in order to do this development and that is also dependent on the time this also depending on how much time i will give size of the device the system the power power will come back we'll come back to power time and again power is a very very important factor in embedded system design nowadays the performance and flexibility okay what do I mean by performance performance means how in in regard to time how fast does it work for example and it can be latency it can be throughput what is the difference between these two latency means the time to take a task let me give an example suppose the camera that we just now saw takes point two five seconds to take a picture right now I have got a camera a some camera a therefore can take four pictures per second right because I have designed it in such a way that it can take a picture with in point two five second process it do everything now so it's to put is also for pictures per second latency is point two five this is latency and this is a throughput now there can be another camera B which has got the same latency but it can take eight pictures per second how is it possible so this is the throughput and this is the latency I said the camera B has got the latency of camera B also as the latency of 0 point 25 second right so the camera B takes a picture in order to take a picture it also requires 0 point 25 second just as camera a date but camera B had two threads running two parallel processing so therefore while one is being processed it could be pipelined one is being processed another is being fetched in that way I can enhance the throughput but the latency is same I hope the differentiation between throughput and latency is clear so that is performance the flexibility is a very another important thing and flicks if I bring a product which has got the flexibility of adapting to suppose now I having a phone which can work in 44 g as well as 5g vasavi there is a phone which is coming in 4G or only in 5g now there is a feature differentiation my phone which I am bringing now and work in 4G and 5g versus your phone which works only in 5g so I have got a flexibility by that I can get a hand over the market that is one another point is another point is that if I have got the flexibility i bring the product in the market and then i find that there are some more demands coming up from the user can i adapt to those demands if my system is flexible i can add new features to the system and make it more flexible ok so these are some of the challenges that we have to face now in order to complete the discussion today will go back to another very important issue the design matrix are all these the energy costs and all those things in any cost you should try to bring down power is another now each of these constraints are competing among themselves they are competing among themselves therefore we have to always find an optimization right there is no unique solution now let us look at this diagram what I am talking of now which is a very vital factor in embedded system design is hardware software raid off and that is why amira system design also goes by the name the task also goes by the name hardware software co design both of these things are designed hand in hand now let us look at this picture here in the digital camera chip we have got so many functions to achieve now this CCD processor preprocessor I can map it to a hardware the JPEG codec I can map it to a hardware or this JP codec had the option that I run a software code on the microcontroller for having the compression right so I have got an option of doing this compression in hardware or in software similarly the pixel coprocessor could be done in a software or could be done in the hardware what is the advantage if I do it in a hardware it will be faster what is the disadvantages of doing it in the hardware the cost the area it will become more it will become heavy all those factors will come into play what is the advantage of making it in software it will be flex I can tune it further but what is the disadvantage of doing it in the software it will be a little slower and there will be the of course I will need a microcontroller and power consumption power consumption will come well deal with that separately similarly this but some of the things that for example the eyes Abbas interface that is hardware so summer some are fixed or assigned esteemed to be hardware some are destined to be software but some can be implemented by either hardware or software and the designers challenge is to really decide on which one should be done in hardware and which one should be done in software ok so we'll take it up from this point we conclude today ok

Transcript for:Overview of Embedded Systems Fundamentals

Transcript for:
Overview of Embedded Systems Fundamentals