 Right now I want the question related to the talk which is the interesting friends in computer networks and talk specific to this take a ways and post workshop homework post workshop assignments. This is the only thing I want to hear about. Anything else please bring it up after the tea break. There is a question regarding this MPLS tagging madam whenever there is a traffic engaging provided by the MPLS tagging there is any another possibility for traffic marting and as well as conjunction control because whenever the over traffic we need to monitor the traffic also. Yes. But this MPLS tagging is provided for only engaging the traffic this is the question related to MPLS tagging. See traffic engineering is not just traffic engineering is lot more traffic engineering does take into account the condition and everything. So, if you have two paths it is not just load balancing you have to determine which path is congested. So, if you have two paths one is giving you 10 MPLS another is giving you only one MPLS. So, when you are loading the paths you will send let us say 90 percent of the traffic on the 10 MPLS and only 10 percent of the traffic on one MPLS. So, you do take into account the congestion levels everything else as part of traffic engineering. So, how it is done within MPLS as I said it is a very complicated topic there are books entire books written on them. So, I mean there is not enough time as I could do an entire course based on that. So, I cannot get into the detail at this stage. So, my question regarding to the potential activities that is you give five options regarding to the create TPS create period instructions like that. Yes. Can you send one of the example regarding these things so that we can refer for the further instructions is nothing but we will take one example among these. So, for example. We can proceed for our real time. So, let me give an example for each so that you are aware of what I am talking about. For example, if you choose the task of think-pair-share Sridhar has provided some examples as part of his slides which I have already shared with you. So, what you need to do if you are using think-pair-share is there is the last slide that specifies how to the interconnection what should be there as part of think what should be there as part of share what should be there as part of sorry pair. Some details are given last slide of Sridhar's talk is very useful and he has also given examples of a particular concept of where he has used that constructor and specified clearly what would be asked in the think-share, what would be asked in the pair-share and what would be done during the share phase. So, what you need to do if you are picking this up is go through Sridhar's slides carefully once more. Pay particular attention to the last slide look at a bunch of examples that are there as part of the slides relate the theory to those examples and then you pick up four different concepts and write the resources based on the constructor. In other words the constructor will have those three fill-in-the-blanks which is the think fill-in-the-blank, pair fill-in-the-blank, share fill-in-the-blank there are paragraphs in fact not fill-in-the-paragraphs. So you fill-in those paragraphs and then you create four such things in document and submit the document to your workshop coordinator. So that is with respect to think-pair-share similarly if you are trying to do something with peer instruction again first thing you should do is watch the video there are slides associated with it go through the slides there is a constructor associated with it just like think-pair-share you have to go through the constructor and then there are also examples given as part of the slides what you need to do is again fill-in-the-paragraphs in the constructor four such paragraphs I mean four such concepts each concept has some paragraphs you need to fill and submit this again to your workshop coordinator. Now if it is a question bank that you are trying to do your reference will be the Bodhi tree platform where you saw lot of this practice quizzes go through those quizzes you know the kind of questions that were asked. So come up with similar questions I would not say exact I mean actually if you come up with an exact question there is no value addition because they are already there you should add on to that particular question bank where by others can use these questions in using as part of mid-sense finals and so on so forth. So you are adding to the question bank the reference there for you is the Bodhi tree platform the kind of questions that were asked as part of Bodhi tree. You could also Google for the text books also I do not mean Google the text books also back of it there are some questions. So you just look through some questions and try to come up with your own questions please do not cut paste plagiarize and so on because it is not going to help because the content is already there you are not doing any value addition there. Then the question is for example animation. So for example or some demonstration of some concept. So you could show some sliding window protocols or you could show the teach CP condition control or you could show some routing all these can be animated again the reference for this there is not I mean some of these text books like I have listed as part of the Bodhi tree they come with a companion website where they do have some of these animations. I will try to put together some such resources and share it with you right now off hand I do not have a reference for you but I will put together some references for you for animation as well and I will share it with you through email. So you look at those references you understand what is it that they are trying to do and do something similar for some other concept. So these are the and of course the lab exercises and project descriptions. So the references for you there are the lab exercise sheets I have shared with you socket programming assignment or NS2 assignment or the TCP dump based assignment or whatever it is. So you use those as a reference and then come up with something so if you are submitting a lab assignment description and if it is a project description I would like to see a description like the description I have shared with you as part of socket programming exercise or if it is an NS2 I would like to see a write up like the kind I have shared with you as part of NS2 exercise. So that is what I am expecting. Before question I really thank you madam for providing excellent material for your videos as well as all the lab sessions. My question is is there any research going on related to the session layer functionality may be related to the session layer function? See the session layer more or less some of these RTP kind of stuff that I was talking about they are kind of managing the audio streams and video streams and so on. So there is definitely some research again if you are trying to do there is nothing majorly novel in that space it is more or less it is an incremental what we would call as an incremental addition it is not something that is shifting the paradigm or so working in that space you can only make an incremental contribution but there is always something or the other happening in that space so you could come up with a better way of adapting to the user band with changing the video supporting better videos how do you do it using some of these sessions session layer kind of a thing. So there is some work that happens but as I said it is the beaten to death kind of a topic so it is difficult to do value addition there. Madam I am having little bit doubt related to the concept you have in your videos related to the hamming on when error rate is high you prefer to use the hamming code but in a single packet more number of bits are in error then whether to prefer CRC or hamming code in a single more number of bits are in error and one more thing we have learned to so many reference textbooks it is always prefer to go for detection followed by the retransmission instead of correction. So regarding I mean I answered this question earlier also where you go for error detection followed by retransmission provided the error rates are small and also provided the you can tolerate that. So you could go for error detection with retransmission if the error rates are small if your error rates are large it is better to go for error correction. So there is again as I said there is an example problem within the Bodhi tree that brings about the contrast between the two. Regarding your first question I did not understand clearly you are talking about whether to use hamming codes or. In a single packet if more number of bits are in error then whether hamming code will handle it. So again it depends upon your end goal so if you want to correct the errors there are plenty of error correction codes hamming codes are kind of a bit of a layman code there are lot more sophisticated codes that you can use to correct the errors. So yeah you can so you will always add more redundant information in a packet and send so at the receiver you could potentially correct based on some error correction code. My question is about sensor network which is connected with software defined network as you have just mentioned while doing the projects student need interface with the different hardware and sensors it is very difficult to build circuits for them in such cases Can you suggest any readymade kits or circuit that can help them to work on software and please put some light on sensor network and their new train. Thank you. Actually I did not understand software defined networks is more in the context of wired networks I did not understand your relation with sensor maybe you are confusing it with software defined radios. So there are two different terms there is something called software defined networks there is something called software defined radios both are very different. So what I was talking about in the talk is software defined networks that is more in the wired network space it has well you could potentially do it in wireless also but right now the focus is on wired space. So I mean I do not that given that I do not understand your question do you mean software defined radios. My question is about sensor network which can be handled by the software so in that case some small kids are needed to interface with the software. So that one has got nothing to do with software defined radios or software defined networks if you are just talking about sensor networks there are a bunch of sensors you could buy sensors there used to be T-Mote I do not know whether it is being sold anymore there are bunch of vendors that sell these hardware and there are a lot of software in the form of tiny OS point the key so on so forth all you need to do is get that software install it and then you could use it to demonstrate many things like routing protocols Mac protocols transfer protocols so on so forth. So the references for this I mean you just have to Google I mean tiny OS and Contiki are two popular operating systems for use on them and earlier I do not know what happened to the hardware earlier T-Mote Sky was being sold crossbow was also selling some but there are enough vendors out there if you just Google you will find out there are enough vendors you just need to buy the hardware download this these are both open source download the software install it and then you are ready to use sensor network. My question is related to software programming if we execute sendUDP.C we are getting IPC4 package and UDP package so is there any way by which we can get TCP fragmentation? See it does not matter that sendUDP.C was a simple program that was written for this particular task that said you could potentially write some other socket programming that will also do fragmentation but typically TCP internally what it does is it determines what the MTU size is and it will put only that fits within the MTU size in other words if you were to open a TCP socket it will not generate a 4000 byte packet on its own because it knows underlying MTU is only 1500 so it will only generate packet size under 1500 that is how TCP is internally hardwired. Now it is possible through some hacking to change that aspect of TCP it is also possible for you to write some higher layer socket programming using what are called raw sockets I do not want to get into those details where you could potentially do many things but as such if you are using a TCP thing it will not generate a 4000 byte packet because internally it knows what the MTU size is and it will not do it unless you hack it. We are selecting the protocol in cloud computing and grid computing environment we are always using TCP protocol but if we choose TCP protocol means each and every time that means each and every packet we need to send acknowledgement it leads to the conversion why we are selecting continuously TCP protocol in this scenario. See TCP protocol as I said has evolved it has been customized for various settings for example if you are using TCP for some kind of grid computation kind of a thing where for example if you are using TCP to interconnect super computers for example you do not use the layman's version of TCP there is a separate TCP variant for that which can make use of effective use of the huge you do not want to deal with let us say 1500 byte packets or once you are using extremely large packet sizes as part of TCP there are a lot of issues with TCP you cannot use very big packets so TCP has been fine tuned for various environments so there is an active ongoing area of research to use TCP custom fit TCP for various environments like grid or even within a data center network maybe you will use some variant of TCP that is better suited for it yeah so does that answer the question so it is not that there is whatever I covered as part of TCP I covered the concepts so these concepts are used and modified to suit the particular environment in which you are using TCP for example there is some something separate for TCP over satellite networks there is some separate TCP I would not say separate TCP it is the same TCP but some configuration parameters are changed to fit the particular environment so my theory is in the current trends of networking as we are using virtualization in broad so this virtualization is carried out in work scale whether it needs a software support or it needs some device support or it needs some sort of any other periphery the idea is so again this is a very recent so for example if you are talking about software defined network it is a very recent topic so there is not much clarity on how to go forward everyone is coming up with some ideas throwing them in the soup people some are implementing some are checking the feasibility and so on so forth so going forward what does it mean there is not clarity at this stage because this is a very active very hot area of research so but at the higher level the idea is you want to use the same hardware and use virtualization to convert the hardware into whatever function you want routing functions switching function firewall function load balancing function whatever it is you want to use the same hardware and do it so that is the high level picture what does it mean that level of detail as I said everyone has their own ideas there is a lot of active research happening so I cannot really there are very many approaches people are proposing whether that is each has its own drawbacks trade-offs that will all be evaluated over the next few years and finally some clarity will emerge or it may be the case that nothing may come out also I mean it is good conceptually it is important also because many feel the need for something like that but then it may not take off also because for example many have felt the need for QoS in the internet for a really long time but only now again we keep talking about QoS QoS QoS but internet really doesn't support QoS so who knows what is going to happen okay ma'am one more simple question nowadays we are listening about big data center network so as we are focusing on data center networking what might be the possible difference between big data centers and data networks big data center network are nothing but data center networks that are dealing with these lacks of force so the network size is significantly larger issues are significantly larger so that is what it is so typically some of these data centers which are quite small which have only hundreds of force thousands of force that is a done deal you know how to manage them they have in fact these shipping container based data centers the whole data center comes in a container you just put it in put it just has some holes outside you just plug in your ethernet cables and all it will do its task if there isn't much that happens there isn't I mean they are being sold as a commodity like a container they just sell it to you just buy it you plug it in you configure it and it is run it is ready to go there isn't much that is happening there it's the big data centers where there is a huge network within the data center how do you tackle the challenges within that is where lot of research is happening good morning ma'am I have a question related to home assignment post home assignment ma'am first question is that the one activity is subject based we have to frame a question 12 question it is subject based from yeah everything I mean I have to emphasize this again everything is in the context of computer networks I don't want you to do some CS 101 programming or database thing per share this is a computer networks workshop and all the assignments are related to computer networks you cannot have something other than computer networks as part of the submission one more question is that one activity is that we have to design a video animation is there any fair where animation software available so typically the animations are done via some I think some java script or some such thing I will have to dig into it a little bit so you may have to write some code for the c code or whatever for that it need not be quite an animation also for example you are providing the c code with some parameters and out pop some graphs that show what is happening that is also fine so I will share with you some reference material for this so that you will have a more clear cut idea so we will break for a tea break