 So where We introduce communication systems very simple concepts in the first lecture. What is a communication system what? That we want to get information from one point to another that's communication and now we're trying to characterize How could we measure how good a communication system works? So our aim is to get information from one Location to another data communications what we looked at towards the end of the previous lecture was different types of information quite simply we looked at some examples of Okay, how big is a web page? How big is a song or a movie and we did a few calculations to finish to to give an idea of Typical sizes of information we want to transfer Towards the end of this topic. We'll try and say well If we want to transfer a movie Photo a web page from one entity one entity to another What does the user require in terms of performance? That's what we'll get to at the end of this topic Which one was biggest those that were here last lecture of those examples, which one was the biggest and remember The movie we did some calculations of an example movie and it turned out that the movie a two-hour movie was about 10 gigabytes nine gigabytes of actually video one and a half gigabytes of audio Let's keep going Sometimes the types of information we we group together sometimes we talk about audio voice calls Radio streaming as example other times video Which usually includes audio someone talks about a video application? It usually means that the the picture and as well as the audio so video conferencing video streams and Other applications or other information types Sometimes just referred to as data Data is a bit confusing sometimes data the word refers to all or any type of information Other times it refers to specific non audio non video types. We'll try and avoid that confusion But there are many applications In the data classification What we want to get to we'll start with today is How do how do we know if a communication system is good or effective? and the three basic measures delivery accuracy and timeliness Okay, a good or an effective communication system will have all three of these as We go through the topic today. We'll see some more details of each of them, especially the last two They're quite simple Delivery means that the data that we send from one point to another should get to the correct destination Remember we want to get information from a to be a sends the data our communication system Whether it's a single link or or the entire internet should get that data To the correct destination be in this case. That's almost obvious that case Accuracy is to say that if a sends data to be Assuming it's delivered to be that the data that is received by be is an accurate representation of what a sent a has some data to send to be they send it what be received should be the same or More generally an accurate representation will look look in some cases. It doesn't have to be identical Okay, so some measure of accuracy Timeliness is that a sends data to be a sends the data now Be should receive that within some reasonable time You shouldn't be delayed too much and we'll talk about how much is too much as we go through these slides an example You need to submit an assignment tomorrow For this course deadline is 5 p.m. Tomorrow, and you need to email it to me. Okay, I'm Getting old and I mainly use email for communications. So you need to email me the assignment By 5 p.m. Tomorrow if you don't email by the deadline then you get zero for the assignment so tomorrow you Do the assignment? Don't worry. This is just an example. There's no assignment Tomorrow you do the assignment and you Finished you've got the answer and you're typing up the email and send me the email Okay, so in there when you compose the email you say to address to Steve at SIT dot tu ac dot th. That's my email address And you type in the message and the answer and you press send now delivery is saying well What happens if when you press send For some reason that email message goes to dr. Tanarak It goes to his email box, but not mine What's your problem? You get zero for your assignment your problem is because I don't receive the assignment by the deadline therefore This has been ineffective communications the communications is sending me an email But if for some reason the system delivers the email to someone else That's ineffective communications. That's all why would it deliver to someone else? Maybe there's an error in the setup of One part of the communication system So that's delivery if we send to some destination it must be delivered to that destination Otherwise, it's ineffective accuracy You compose the email and in the email you must include the answer for the assignment and You type in the nice email. Here's my answer to the assignment. The answer is 42 that's the answer and you press send and you send the email and It's delivered across the internet arrives at my email inbox and the text says here's the answer the answer is 43 The received message is different from the sent message again. You fail because you gave me the wrong answer So if for some reason the data received is different from what was sent Because of usually a failure in the communication system then it's ineffective communications So accuracy is making sure that what's received is Accurate from the receivers perspective Sometimes it needs to be identical in the case of the email. It should be identical if you compose an email and press send The email I receive should be identical to what you typed in If not, it's not a very good email delivery system But in some cases we'll see it doesn't have to be 100% Timeliness you send the email tomorrow at 4 p.m. The deadlines 5 p.m The internet delays this email and it arrives in my email inbox at 6 p.m. You get zero again. Okay, so The communication system must deliver the data in a timely manner in within reasonable time What's reasonable for an email anyone From when you press send and to when someone receives the email, what do you think is a reasonable time? How long should they wait? Yeah, it depends on size, but how long would you wait? Yeah Two seconds a few seconds maybe Okay Maybe an easier one Let's say you're using Some instant messaging application old days MSN line or similar where you type a message Press enter the other person sees it and they immediately respond or immediately What's the time? What's reasonable time from when you press enter until when the other person receives the message? At most ten seconds, okay, what if it was one hour? You're using line you press enter and they receive it one hour later in effective communications one second, okay Milliseconds, okay, so Different applications will have different requirements of what timeliness is good Similar what accuracy is good towards the end of these slides will give some more examples of them So that general measures of what is effective data communications We're going to have some more specific measures in a moment, but before that what of what is this? again, sometimes applications that transfer information are classified by different names or different groups in In terms of the internet applications we use in the internet Sometimes we refer to traditional or normal internet-based applications and And other types being multimedia or real-time applications some examples Internet-based file downloads email web browsing instant messaging Sometimes remote logging connect to another computer databases In these types of applications accuracy is usually most important again Yes, you type in an email you press send The data received should be identical to what was sent accuracy should be 100% you Visit a website, which is really downloading a web page from a server to your browser The web page that your browser gets should be the same as what was on the server The accuracy should be 100% with multi-media or real-time applications like audio or video streaming watching YouTube Maybe more so Interactive applications like voice or video calls Skype or similar gaming Collaborative desktop sharing applications in those applications Timeliness isn't usually more important You are talking using Skype or similar using your computer or phone From when you say something on your device Until when the other person receives the audio representing what you said the delay should be quite small Otherwise if the delay is in the order of a few seconds you speak and Then there's a delay before they receive it and while they're waiting for to receive and hear you speak they start speaking and you start to overlap in your talking and Most studies show that delays in the order of several hundred milliseconds is usually desired gaming if you play online games you Multiplayer games for example you see the screen where you see other players which are nearby And you press the button to shoot someone else Then from their perspective the other player they should get feedback of what happened almost immediately If there's a large delay from when you do something until when they see something the game will not work very well People talk about a lag in the game That's too much delay or not timeless timeless delivery Email web browsing can usually tolerate more Delay or less timeliness delivery than these applications Whereas these applications and we'll get to it later can tolerate some less or lower accuracy when you're Streaming a video you're watching a video on YouTube There's a video on the YouTube web server and They send it to you to your browser and the browser displays the video If you don't receive 100% of the video you can still play it back and watch the video What may happen is that you start to see some small artifacts in that in the video on your screen that is You're watching the video. It's high resolution high quality But some of the data doesn't get received by a browser most video Formats and players are smart enough to handle less than 100% accuracy of the data delivery that is Imagine you send in the server sending the video some of the data sent doesn't get to your browser in That case what happens? Well, maybe some of the pixels in the video are not displayed correctly on your browser but if a few pixels out of Five million pixels are not displayed correctly for a fraction of a second then you usually don't notice that So and we'll see some more examples of this as we go through through this lecture and even the course With multi-media applications accuracy Doesn't have to be 100% sometimes the data received doesn't have to be the same as what we send What about delivery? Three measures delivery accuracy time limit us delivery in both cases needs to be correct Delivery is usually you have it or don't We both both cases need correct delivery Let's move forward so we can get some examples so our Three effective measures of communication are quite general delivery. Yes, or no, we have it But accuracy what how do we measure accuracy? How do we measure timeliness and how do we measure other aspects of communication systems? We'll talk about performance metrics Ways to measure the performance of a communication system So a metric and we'll go through some example metrics some common ones that you probably already know about There are others, but we'll go through ones which I think most of you will know or have seen in some examples We use metrics So a metric is a way to measure the performance of a communication system We'll look at different metrics Why do we use them? How do we use them? Sometimes we want to measure the performance of real systems. We build a communication system. How does it perform? So we use metrics to compare to see how it performs Well, sometimes I want to predict or estimate what the performance will be of a new system So we use metric metrics to do that so you get a job and your job is to Choose the communication system for your company For a particular purpose. You need to choose a technology and one way you do it is compare based upon some of these metrics Other things will be cost and other factors. We represent the different metrics usually using different statistics But let's come back to them as we after we go through a few of the metrics themselves I'm about to show you on the following slide some example metrics, but there are others Some of the others will see during the later lectures Let's go direct into the simpler ones Data rate is a common thing that we will refer to and we'll see When you need to select a particular communication technology. I'll try and give a simple definition of these metrics Maybe some other names which you may have heard of which mean the same thing or a similar thing and some examples By data rate, I mean the rate at which the data the information is delivered from one point to another So it's a rate think of a speed Data we will normally measure in bits bits or bytes If I use bits you can easily convert the bytes vice versa So the data rate will use as a base unit as bits per second How many bits per second can we get from one point to another? I didn't bring my land cards today, but those that were here in the last lecture. Remember the old land cards My land card in my old computer at least and the ones that I had last week could send the technology Built into those land cards could send 100 million bits every second So when you buy the land card, that's a characteristic of that hardware That it transmits at a maximum speed of 100 million bits every second quite simply The data rate would say is 100 megabits per second So it's a measure of how fast our device can send our data our bits Can anyone remember or know the data rate if I send wirelessly from my laptop to the access point on the wall? No, I want to have a guess Maybe check on your phone It may even tell you the data rate when you connect to a Wi-Fi access point. You want to know the typical values How fast can I send that access point on the wall? What's the data rate? It's an old one if you can't see it. Is that the answer? Okay, anyone want to make a guess 100 100 watt megabytes per second All right my The typical speed that you get when you plug the LAN cable into your computer is either 100 megabits per second Or if you're lucky 1,000 megabits per second Typical depending on where you are and how old your device is Do you think Wi-Fi is slower or faster than wired LAN? Hands up for faster Good good hands up for slower Wi-Fi is slower than wired generally Okay, that's not always true, but generally this the similar Type or cost of technology wireless is slower than wired This old ish access point supports a data rate of a maximum 54 million bits per second 54 megabits per second my old wired LAN 100 megabits per second wireless 54 about half But newer ones can support faster Up to several hundred megabits per second Similar that newer LAN cards can support one gigabit per second 1,000 megabits per second They are the What we call the data rate the rate which we can send bits across a link normally Anyone know of some other examples who has home Internet in their dorm not wireless, but ADSL or cable anyone Anyone who's used it before Has anyone used the internet before just just check if people have pans they can put up Sure, you've not used the internet Okay, it'll be hard subject Okay, if maybe if you've used home ADSL Okay, or your parents or your friends have ADSL ADSL is a technology for Commonly used for internet access inside homes Where you use the telephone line? The data rates which the modem supports usually in the order of several megabits per second The maximum around here is around 24 megabits per second For example at home I have a plan that I pay I don't know 600 baht per month and The data rate at which I can download is a maximum of 10 megabits per second Upload may be different anyone have cable cable internet and how fast 10 Usually you you pay for faster as as normal you but generally cable can go faster But you need to pay more Any people from outside of Thailand here and he maybe have different technologies or different speeds How fast is home or internet in the dorm or in an apartment? examples Ten around ten megabits per second Yeah, okay, so in the order that's around the tens of or few to tens of megabits per second, okay let's look at Some other examples of technologies that you may have come across and I'll use Where I just grabbed some or go to Wikipedia and I'll just grab a few examples of different technologies This is the list of device bit rates Another name. I will call it a data rate, but sometimes called a bit rate Some of them you'll know of Before we go to it who uses internet on their phone Okay Not Wi-Fi, but what 3g how fast 3g 42 what Mega bits usually not always the case, but usually data rates are measured in bits per second not bytes the B means bits You buy a hard drive and it's usually measured in bytes you buy an internet connection or you pay for it It's usually the B means bits per second. Okay, not always but common. So some 3g or three and a half g why wireless mobile phone data rates are in the order of 42 megabits per second 14 megabits per second was an older one, but again you may pay The faster the speed the more you pay Let's find a few this is a quite a large web page with many different examples. I'll just go to If you can't see it You can have a look in your own time. Just Who did their homework? You still got some time to do the homework the homework is simple from last week visit the website log in and Make sure you're aware of how to use the website and even try one of the practice lessons there The first practice lesson talks about bits bytes megabits megabytes giga and so on the different prefixes and different units that we come and commonly come across and These numbers are an example and Just quickly see the the lesson to get further explanation, but when I say mega I mean 1 million as here One followed by six zeros mega one million in some forms of computing We use binary prefixes For example kilobyte is not one thousand, but one thousand and twenty four When you talk about hard disks and file sizes often One thousand and twenty four is meant by kilo, but I will not use that in this course It's much easier for me to calculate in terms of thousands millions and billions To see that explained see that try the lesson Let's go through These are old ones remember old-style modem how fast Dial-up modem the strange noise modem Sorry, there's a lot of information here. I'll just select a few of them the typical speeds the normal speeds For the dial-up modem Around 56 kilobits per second If you had that dial-up internet access 56 kilobits per second 56,000 bits per second some other ADSL to There are different variations of ADSL for home internet access ADSL Maximum speed because it can be lower than this 24,000 kilobits per second or 24 megabits per second That's download when you download from say the internet to your home Upload is limited to about three and a half megabits per second. So there's different speeds up and down Cable modem if you have cable internet access a common one around here is What's called doxis version 2? 38 megabits per second down 27 megabits per second up There's in fact a doxis version 3 which goes up to 160 megabits per second down So doxis or cable modem and ADSL are commonly used for home internet access or business internet access Many other technologies there. Let's just skip through to a few that we may You may know mobile phones many different ones here Edge maybe before 3g you may have been using edge which Was the order of hundreds of kilobits per second internet access to your mobile phone? 3g There are different technologies here's one HSPA Three and a half g listed here 14 megabits per second So just some examples of data rates of different technologies And there are extensions up to 42 megabits per second for for 3g sorry control and Some of the later Mobile phone internet technologies go up to 300 megabits per second 4g or LTE is sometimes the name What about for example, so that are the things that we may use mobile phones home internet Wi-Fi is listed here. We'll see it later, but we'll see some different ones What about that companies may use to connect between officers between cities? What about when we connect say from Bangkok to Singapore for the international internet access? How fast do you think? Let's see some examples. These are connecting between cities or between countries sometimes we refer to as wide area networks a Network across a wide area The older technologies with the order of one two megabits per second still around some of them We go down OC means optical carrier which is referring to optical fiber and the different Standards of optical fiber common ones OC 192 10 gigabits per second. Okay, so International links commonly using optical fiber in the order of tens of giga 10 gigabits per second and Up to maybe 160 gigabits per second for latest technologies. That's enough for them Lands inside an office or a home Where do we get to and down the bottom? I hope wireless lands this Wireless access point is 802.11 g 54 megabits per second But newer ones go up to 11 n 600 megabits per second and the latest Technology released is AC up to seven gigabits per second under under very special conditions Have a look through there just to get some examples of different data rates sometimes called Bit rate the rate at which bits are sent across a link Capacity that is how much what's the maximum we can send across a link What's the capacity of the link and we'll see a few others that may come up in the later lectures? similar names for data rate delay another metric this comes back to our timeliness measure The time it takes to get data from one point to another Also called latency The latency is equivalent to delay Delay we usually mean to get from one point to another but sometimes we're interested in to get from one point to another and then back Okay, to get there and back to get there and back is called the response time or sometimes the round-trip time Response time send a request get a response. What's the total delay round-trip time time to get there and Back make a trip all the way around measured in seconds Two quick examples I send an email at 10 a.m Press send it arrives at the destination at 10.03 the delay is three minutes The email delay is three minutes I'm web browsing. I click on a link at Time 1.4 seconds. Okay, if I start the clock then and And at time 2.6 seconds so 1.2 seconds later I receive the web page displayed on my web on my browser Then we'd say the response time is 1.2 seconds Because we covered in the first lecture web browsing is you send a request to a server for a web page The server sends back a reply a response So how long does it take from when you send the request? Until you get the response That's the response time in this case. I send the request at time 1.4 I get the response at time 2.6. So the response time is 1.2 seconds the difference between as anyone Come across delay latency a response time or round-trip time in any other applications Does anyone ever played an online game? Maybe a multiplayer game you may have heard of ping time The time it's some measure used to to measure how good the network is to play that game because again You want a quick response time the ping time is really a measure of round-trip time Time to get to some destination and back maybe to a server and back We will look at some more detail of how how do you calculate delay or what contributes to delay? There's a few slides later for that one. What else? Error rate Something you may not have seen so much but now a measure of accuracy our two general measures of Timeliness and accuracy timeliness we can measure by delay accuracy by error rate sometimes I Send data from A to B B doesn't receive all of the data How much didn't it receive? Well, we can talk about well the data not received That means it's an error. It wasn't delivered What rate of errors do we have out of the total being sent? So error rate I'll define as the fraction of data sent That doesn't get delivered to the destination We want it to be low Sometimes it's the bit error rate. I send a Thousand bits across a link on average 23 bits arrive, but they don't they're in error What does it mean if a bit is in error? I transmit a bit one I Transmit a bit one it arrives at the destination as a bit zero That's a bit error When I transmit a bit one when it arrives at the destination the destination should interpret it as a one That's successful delivery if it arrives is the opposite bit and it's an error So if for example, I have a link where I send a thousand bits and on average 23 of those bits arrive in error They arrive But they are wrong and we could say the bit error rate is 23 divided by 1000 23 out of a thousand or 2.3 percent or as a fraction zero point or a decimal 0.023 There are no units here. It's a ratio or More generally I send an email to all 100 students in a group Something goes wrong. Five students don't receive the email. I can say there's an error rate of 0.05 five out of 100 Or five percent We'd like the error rate generally to be low Okay, we're getting into some newer ones here that you may not have come across But I think not too hard to understand Actually before we go let's go back a step and look at some examples of delay How do we measure delay in a computer network does anyone measure the delay before now Play a game and then Specifically in that game. What's the term on or what's the thing that measures the delay? lag is sometimes used another name you may see is Ping okay ping refers to actually an application the concept of peeing something is at least in computer networking is send a message to see if it's there and That entity will tell you It's there by saying a response if you get a response, you know that you can communicate with that entity But in terms of delay ping is we send a message to some destination The destination when it gets that message sends back a response The time from when you send the message until you get the response is called the ping time or the response time or round-trip time I have a Program called ping on my computer. We can test the round-trip time for different locations. Let me get this correct Let me just set up and make sure we have our network working. It's slow Okay, be patient I have my computer and I'm going to use the ping application just to test the delay or the response time from one location to another To do this demo I'm going to use my computer in Japan. Okay It doesn't work inside SIT okay the SIT network does not do not does not allow us to do the things that I'm about to do So I'm actually Remotely connected to my computer in Japan and we'll do it from there. So everything I do imagine we're sitting in Japan at the moment We'll use ping to test the delay between my computer in Japan to other computers in the world Don't worry about the commands. I'm typing. It's not so important, but I'm going to ping the Google web page in Japan a lot of information shown I'll just highlight what we care about What happened? I? pinged from my computer in Japan to the Google website in Japan the idea was to send a message to the Google web server and That would send back a response and I did it five times a count of five send first message Get a response and the first response Focus ignore that most of this focus on time the time from when I sent the request and Till I received the response was 2.13 milliseconds That's called a ping time the round-trip time or the response time It's time to get there and back and then My application sent a second request and got a second response the time was 1.8 milliseconds and then the other three requests and responses Because I specified to do just five the count of five and It gives us some summary statistics the round-trip time RTT Minimum average maximum and the mean deviation the average 1.865 milliseconds roughly it's about Close to two milliseconds from my computer in Japan to the Google web server in Japan It doesn't matter about the command or all these details just take note of the time Let's try a different one a different destination from Japan to the SIT web server Anyone want to estimate what the time will be? seven hours Seven seconds seven milliseconds Let's find out about a hundred and fifteen hundred and twenty milliseconds okay, so My computer's in Tokyo SIT web servers in Bangkok So what's happening? My computer sends a message Tokyo to Bangkok Takes about 60 milliseconds 60 milliseconds to get there and Then the server responds about another 60 milliseconds to get back I estimate meaning a total of about 120 milliseconds. Remember. This is the round-trip time The one-way delay Would probably about half of that Okay About 120 milliseconds. Why is it much much larger? Japan to Google Japan two milliseconds Japan to Bangkok 120 milliseconds. Why? It's further away so We will do calculations later and see how we could calculate delay in ideal conditions and People are right and it's further away One aspect of that is the signal That represents the information must travel a larger distance Remember how fast a signal travels a signal like light Travels at the speed of light the further it needs to travel The longer the time it will take so one aspect of the delay is how far physically it must travel In the physical distance, but there are many other aspects as well Like the speeds of the links like the speeds of the computers one or two more Japan to The UK okay, so here from Japan to the UK I Don't know which direction it takes across the world Does it go across the Pacific across the US or does it go? West across Asia, I don't know but it takes about 240 milliseconds in that case much larger distance But there may be other factors in play You can test this maybe from a computer at home using ping it works in Windows it works on a Mac It doesn't work so well inside SIT, but at home it will any questions Questions so far. How's time going 30 minutes? Everything okay first lecture Okay, good Do I have an example of error rate? No not yet overhead in many cases We introduced in the first lecture. We use protocols to communicate protocols to find how we send our messages In many cases to work correctly We can't just send the original data as is We need to send some extra information to make it work correctly That's that's hopefully he's successful here Good everyone's comfortable for the last 30 minutes We survive anyway Overhead The amount of additional data Needed in order to deliver useful data Now we're going to distinguish between two types of data. I want to get I Want to get a What's an example? I want to get an I want to download a file. Okay, the file is one megabyte So the file is on some server And I want to get that one megabyte to my computer We mentioned in the last lecture. We don't just send the file as one large piece of data one large Megabyte we may break it into smaller chunks into packets and send them one at a time and The protocols involved in doing that every time they send part of that one megabyte They may add some add some extra information to keep track of those chunks Let's say I split those one thousand one million bytes into one thousand packets one thousand chunks Then when I send a chunk or the server sends a chunk it would include extra information identifying This is chunk one This is chunk two chunk three and so on it may include a sequence number in there What else may be included Every time we send a message the message may include the address of the destination Last lecture we mentioned addresses like URLs IP addresses To make sure we get effective delivery to the correct destination when we send our data The message may contain the destination address So it's very common that we send some additional data With the the real data. We want to deliver the real data is that one megabyte file, but I attach some other information Overhead is how much other information? so it's a measure of What extra do we need to make this work, but it's not the real useful data Bit hard to give Too many examples yet until we get into some real protocols later, but a couple here is Measured in bits so the overhead how many extra bits do we have to send? Some of you know about parody checks may be in computer hardware error detection schemes For every eight bits of data that I have I may add a two-bit parody check to make sure that Someone who gets that data can detect if there are any errors in that data It's used in computer hardware or in file systems to keep track of files, for example So there's eight bits of real data Plus an extra two bits to make sure that the data is delivered correctly We say those extra two bits of the overhead another example I send a chunk of data as a packet containing 1,000 bytes of real data and 50 bytes of additional data Often referred to as a header It's usually added at the start at the head of the real data The overhead is quite simply the extra additional data of 50 bytes So it's the extra stuff we need to send to get the real data to the destination Why do we need it? Well, we'll see many different cases to include addresses to include sequence numbers to do error Detection if something goes wrong for the receiver to be able to detect if there's errors We'll see those through the course Let's get to one more and then to through well throughput here. We are Data rate the rate at which the data is delivered from one point to another For example the data rate that I can send bits from my laptop to the access point is 54 million bits per second throughput is The rate at which useful data is delivered to the destination Let's go direct to some examples to explain I think You may know when you download a file from a website and the browser gives some feedback about the bits per second or the time it takes It may go up and down The download speed may vary over time and it's usually much lower than you expect or you hope Because often we know about the data rate the data rate of Wi-Fi is 54 megabits per second of your mobile phone 14 megabits per second of your home internet 24 megabits per second That's the data rate the rate at which we can send bits But there are overheads that always occur Plus there are things that may go wrong So another measure of performance is what's called throughput. It's the rate at which the real useful data gets delivered and It's less than the data rate Let's see if we can get an example I'm going to download it. I'm going to download a file from my computer Not in Japan anymore, but back to here from my computer from Where from the ICT web server? Okay, our course website is on the ICT. S I T T U A C T dot th Where is that web server? Anyone know it's a computer somewhere? Where is it? ICT Third floor of this building. Okay, it's downstairs a few few floors. It's just a PC sitting in there That's the ICT web server Let's download a file from it and see how fast it takes to download I've put a file there that I can Copy I think So W gets just a program that would download the file for me the web the address ICT web server domain and the file I created before is just some random data. Okay, it means nothing See if it works. It's downloading The file was 10 megabytes and This software downloads the file. It was 10 megabytes the file and it gives us what some summary Maybe here 2.36 megabytes per second. That's the throughput It is a measure of maybe Here there's maybe some approximations or some different calculations going on here that the timer is not that accurate 10 megabytes in 4.0 seconds Equates to 10 million bytes divided by 4.0 seconds 2.43 million bytes per second That's the calculation of throughput. The real data is the file The real useful data is the 10 megabytes the time it takes is four seconds So the throughput is simply 10 megabytes divided by four seconds or 2.43 megabytes per second million bytes per second Let's download again same file faster The throughput in this case it took 2.2 seconds Why? Same location the servers still downstairs same file same size Why I don't know the exact answer why? But note that it varies so throughput may vary under at different times the reason why in this case most likely is because in a computer network The speed depends not just upon you downloading, but what others may be doing at the same time Someone else may may have been downloading a file at the same time as the first case Therefore mine had to wait a little bit. It was slower. Whereas in the second instance Maybe no one else was downloading from the ICT web server and therefore mine was much faster I don't know unless I analyze that in depth, but the throughput First is the rate at which the real data is delivered It may vary over time and in fact is hard to predict how many megabits per second approximately of the faster one Megabits per second easy one How many megabits per second? Approximately 35 okay, so four times eight. This is note the uppercase B means bytes Lowercase B normally bits. So this is 4.4 megabytes per second times by eight so around 35 megabits per second I'm using my wired LAN, okay Downloaded by the wired LAN. Let's see if I can I'm going to add part unplug and If I can and switch to Wi-Fi Let's hope this works Now I'm downloading and I'm using the wireless link same file 10 megabytes took seven seconds 7.1 seconds 1.45 megabytes per second why is it slower? Why is it slower? Wi-Fi good Most likely again, we could we cannot know for sure but I guess that When I was using the wired link The maximum speed I could send was 100 megabits per second That's from here to the next device and then there's another device another link to go via When I'm using Wi-Fi the maximum I can send to this wireless access point here is 54 megabits per second But I know that there are many overheads involved Using Wi-Fi even though the data rate is 54 megabits per second the overheads practically mean The best you can get is about half of that and Later we'll do some calculations as to why but with wireless Best you'll get about half and even worse if many people are on their phones using the same Wi-Fi access point at the same time my Throughput will be even lower So it depends upon what others are doing So I guess that the reason this is slower is because the of the wireless access if I plug the wired land back in it Would go up again Let me go back to wired and do two more examples Sometimes these don't work as I plan, but we'll try Downloading a different file from Kazat Sart University. So also in Thailand So this is from KU Kazat Sart. So the other side of Bangkok I'm about three megabytes per second. So this is Before it was to a server downstairs now. It's to a server on the other side of Bangkok. Okay, 20 30 kilometers away About three points or 3.6 megabytes per second Doesn't matter what the file is but I know that there's the same file on other servers in the world I'll download the same file from somewhere else This one's in Singapore Same file. Let's hope it connects There we go So this is from us to Singapore Well Singapore sending the file to here in Bangkok in fact, so You'll see it's it's going down. It's around. Oh, okay It goes up and down up and down Okay Many different factors impact on it here We averaged about one and a half before We got three and a half more than three and a half megabytes per second last one Same file, but in Australia ETA three minutes. Okay, we may not wait till the end Why? Why is it different? ping Okay, so one what's the difference? So we had Bangkok to Bangkok Bangkok to Singapore Bangkok to Australia actually the opposite direction Why is it what's different? I think most people can recognize the physical distance is different The delay may be different But many cases the delay doesn't matter the time to download a file once you get started Once the download starts doesn't depend much upon the delay Or the the distance it depends upon other factors many other factors. What's one of them? This one's still going. Okay, what what could be another factor that impacts upon the throughput anyone? Okay, if other people are accessing say the server at the same time the server needs to respond okay, so In many compute communication systems that the number of users using it at the same time Means the system slows down What else? Think about all of the links between my computer and the server the first link using white wired access is this cable going From my computer into the wall it goes up I'm pretty sure it goes direct into a device in the third floor and then that device has a cable direct into the server There are just two links between me and the server Sorry, this was a server in Kazat start. So there's another link. There's more links inside Thailand Whereas going to Australia the links are inside Thailand and then probably Australia down to Singapore and then most likely a Submarine cable Singapore down to Sydney or somewhere Okay, there are multiple links and then inside Australia to the destination The speed of those links may have a large impact upon the throughput the data rate of those links Generally if you have a sequence of links that you must use Your limit is the minimum of the data rates of all of those links Sometimes called the bottleneck link if my data passes through three links one is one megabit per second the second is 10 megabits per second and a third is 50 megabits per second the best that I can get is limited by one megabit per second link So the set of links and what are their data rates will impact upon my throughput today, we're just introducing these concepts over a A few of the lectures will see some more calculations of how to how to calculate throughput and Some more details of what would impact upon throughput Did we get to the end in a few seconds? Yes so Like the examples I just illustrated If I was downloading a 12 megabyte file from a website and it took 26 seconds Calculate the throughput how? Throughput is the rate at which useful data in this case the file is delivered to the destination and Here we'd really talk about the average throughput on average if we Take 26 seconds to send 12 megabytes on average. We're sending or receiving at 6 million bits per second how just File size divided by time We don't need that anymore What do we have? 26 our file was 12 megabytes which is Let's convert to bits Times by 8 is 96 megabits Or 96 million bits That's how much real or useful data. We have to deliver it took What do we say 26 seconds To download so we can say the throughput the rate at which That useful data was delivered. We're delivering 96 million bits in 26 seconds, which is whatever it was on our lecture slide. What was it? Six megabits per second approximately is that right? I don't think that's right is it? Doesn't sound right. What did I do wrong? 3.6 megabits per second Good 96 divided by 26 is three and a half or 3.6. Correct. Why is my lecture note say six? Because I made a mistake somewhere You can be better than me and use a calculator about 3.6 Megabits per second fix your lecture notes So it's just the total size that the total number of useful bits divided by the time it takes to deliver them the rate This one's easy to calculate if we know the total size and know the time it's just this division But in many cases in some networks, we may not know How long it takes we need to estimate what the throughput would be in advance So we need to look at overhead, but let's stick with this simple one for today Any questions on this calculation fix your lecture notes fix my error in there 3.6 not 6 We'll come back to it some other calculations later about throughput Do we have anything to finish? efficiency efficiency is related to throughput and overhead in general the fraction of time we spend using a resource for its intended purpose Come direct to this example Let's say my Wi-Fi link has a data rate of 54 megabits per second I know that when I buy the laptop and the access point the specifications say 54 megabits per second and when I Usually I can select and see. I don't know if it does this connection information Sorry, I'm on my wired link Ws it and I'll get there I can unplug Speed is at 2 megabits per second 5 megabits per second for my Wi-Fi up to 11 megabits per second The maximum it will go to is 54 here. It's not going to go there for me today because It changes depending upon the signal maybe and how strong the signal is for my laptop But if we have a data rate of 54 megabits per second But I download a file and I measure the throughput to be 20 megabits per second Then I could say the efficiency in that case of using the link is point three seven 20 divided by 54 you new check if I've got that calculation correct My data rate is the maximum speed at which I can send bits The throughput is the true speed at which I deliver real data So the efficiency is well what fraction of that maximum speed do I utilize? 20 out of 54 37% so in that case I'd say the efficiency of the link usage is 37% We'd like that to be as high as possible So this is just 20 divided by 54 another example Let's say I pay a thousand bar per month for my 10 megabit per second home internet access On average every month. I download at two megabits per second. Well, I'm 20% efficient I only use that resource on average 20% of the time We will use efficiency to compare different systems and to compare different protocols I think we'll stop there Next lecture when tomorrow we'll look at I'll try and find a few more examples of those metrics And then we'll look at delay come back to delay in a bit more depth do some more calculations