 Hi everyone! I will be talking about routing protocol in wireless mesh network. This paper is not technical, it's more of conceptual so I hope all of you will surely go back something new after this. This was written exactly 10 years ago from Canada, Waterloo University and I will be working a lot more on internet of things technologies quite deeply in the next year to come. So I thought of revising this thing and this paper in a way is a really coming full circle for me and I can only connect the dot backwards to exactly 10 years ago. For my undergraduate in electrical engineering my final year project was on position-based routing protocol for mobile ad hoc network. So let me talk a little bit about it. So my professor from I2R where Brahim works and used to work, DSI isn't part of I2R? Okay sorry, used to work and Huixian also graduated from I2R Research Institute some of you might have heard. So McProp gave me this problem like how do ships that are parked at the Singapore's near the coast but it's about 20 kilometers away from the coast get internet? So how do ships get traditionally internet from the satellite right? And what you it's very expensive you really have to go there and come down and if you have relatives or friends working in the shipping industry and they go sailing out there you don't have internet. So my problem basically was like using this wireless network called Y-Max maybe some of you have heard of it Y-Max. We are gonna make each of the ships as a node and using point-to-point multi hop it's gonna come back right to the shore and communicate with the cell tower there. So at least the ships you know at the coast of Singapore will have internet instead of communicating with the satellite. So my problem was to device a routing protocol and that will be basically location-based using the latitude or longitude and you know all the ships are moving and that's why it's called MANET mobile ad hoc network. So I did that project okay and I graduated and I was okay that's it throw away and but guess what it comes back and bite me again because I have to study a little bit in depth of about what wireless mesh network is and today wireless mesh network it's just coming up if you can see your home network you will hear a new technology such as Zigbee or Z-Wave coming up and these are basically mesh network as well. Sure they are not wide area like the ship they are within a room or within a building. So that's what I decided to read this paper and go through some of the concepts to consider when designing this. So I will start off with the topology and I think this is something all of us will understand there are two different there are many different kinds of topology but the ones that we are familiar with is star topology. So if you talk in terms of Wi-Fi routers of the hackers face as a Wi-Fi router and all of our laptops or mobile phones are basically connecting to it. So it's like a star so the node at the center of it is the router and similarly our mobile phone as well right so we have all the mobile phone and then with the wide area network the cell tower is basically at the center of it. Bluetooth and satellites is also similar it's star topology but mesh network is something different they basically talk to one another there's not such a central thing so Zigbee and Z-Wave can be configured as a star topology but it can also be configured as a mesh network and so is Laura Van so it can be configured both as star and mesh. So there is a value in having mesh network so what is the advantage of having wireless mesh network. So the first thing is the last mile connectivity so imagine today in a remote village in Africa from the city if you want to deploy internet connectivity to the villages the last mile you do not have to install wires so you can just connect access points and then you will have internet. It is also low cost of deployment because you do not have wires or fibers to implement also rapid deployment just make the access point go there it can also be self-organizing each of the node self-optimizing fault tolerant you can use it to set up ad hoc networks and then again bring it down again set it up again and also sensor networks. Sensor networks are environmental monitoring networks and that's something I think I'm interested in. So these are basically the advantages of a wireless mesh network so it's called WMN it is different slightly different from MANET which is Mobile Adhoc Wireless Network so they're slightly different in wireless mesh networks the nodes are not moving whereas in MANET the nodes are moving. So let me talk a little bit about what are the challenges in wireless mesh network as compared to the traditional network that we are all aware with. Firstly point-to-point multi-hop so you are not directly connecting to the router you have to kind of hop and then go to the final destination. Using this there's also a high demand of quality to ensure that whatever packet you kind of initiated it must reach the destination. So I want to talk a little bit about the different types of mesh networks so firstly so and the paper talks about three different types firstly the mobility the power and the traffic pattern. The mobility is quite obvious each of the nodes is moving like my final year project the nodes were ships and in future you might have cars moving and they might like communicate with one another on the road for let's say some kind of autonomous or driverless safety detection algorithm. So in this case all the cars are moving but if you have some kind of sensor nodes they are kind of fixed they are not moving so in this case the design considerations will change. The second thing is power are each of the nodes having power or not. For the case of car yes it has power if it is a sensor network it might have power but if you know it is a low power powered by battery it might not have power so in this case once again your routing protocol has to take note of that you cannot be sending I don't know huge amount of data and then at a very high frequency you might drain out the battery. The next part is the traffic pattern where exactly is the data going to is it just going to the next node and communicating with one another or in the case of ship it's all going to the cell tower for connectivity. So these are three types of things to consider in terms of mesh network. The typical mesh network architecture has these three things and I think this is quite similar to maybe our mobile cellular network as well there's a network gateway which has an access to the wired internet the big internet it might or might not have that there's also access points like this these are assumed to be static they are not moving and they are also assumed to be a very fault tolerant and they have unlimited power which means they have connection to the power. The mobile nodes in this case can be your mobile phone can be sensors moving around and they can vary in computation and transmission and power. I thought this is not from the paper but the paper did speak about different types of networks so when designing mesh network you have to consider the area that you're trying to cover. So there are four different types one is personal network and I think some of you how many of you have already have Fitbits and you are communicating with your mobile phone it's basically around your body and that or even our MRT cards are what RFID is it or NFC. So RFID or NFC are even shorter range right when barcode as well like short range barcode is not electronics but yeah so RFID and NFC is even shorter then comes the area the personal area network any personal area network you have the Bluetooth low energy as well as Bluetooth and then you have a WLAN which is the wide area network and this is where 802.11 which is the Wi-Fi and then if you go slightly bigger this is where you will have the urban infrastructure coming in this is like beyond our room and that's where you have things like WISAN or Zigbee. Now Zigbee has a lot of different types of implementation one of them can be inside your room for smaller area the other one can be a wider area and of course the large one is the wide area network which is the cellular network how our mobile phone works outside our rooms. Yeah so there are different different implementations so Zigbee I I think it depends on the physical layer maybe Huixian can correct me if I'm wrong but yeah but yeah but they have very different types of implementation how you implement Zigbee. They have lower frequencies Zigbees that can go for multiple kilometers. Yeah so Zigbee is not just one type it can be configured in very different ways and of course so you know our cellular that we use with our mobile phone is basically for high power because our mobile phone if you notice we have to charge it every day right but in future there's something coming up that's called low-power wide area network and in many countries in fact I read the news recently like South Korea is implementing it's called LP LAN LP WAN and you will hear this term more and more so whenever they will say things like oh you know this country is implementing a IoT network which is different from your cellular network it basically means that they will support low-power wide area network which means they have low bandwidth because it will just send you very tiny amount of info not like your mobile phone it will send voice and data and it will be wide area just like your cellular network as well so the LP WAN is something you will keep hearing about and I believe Laura WAN is something that is open whereas the rest of the implementations like SICK Fox in fact Hackerspace had a SICK Fox project a couple of years back SICK Fox or others are not our proprietary but it seems that they are all based on the physical layer 802.15.4 physical layer so all the WP WANs have the same physical layer and they basically have proprietary layers on top so I feel this is very important for us to segregate like you know like these are the different kinds of network coverage all right so next what are the different characteristics of a wireless mesh network so there the paper talks about four different characteristics transmission medium the way you deploy it to the technology and the infrastructure so in terms of transmission medium you definitely have to take care of the bandwidth you know like Wi-Fi is high bandwidth whereas BLE is low and and they're always trade off I remember Kingman if you know that guy but anyway he showed me once this never mind you don't need to know him he basically showed me this wireless triangle of like you have to sacrifice in one of the ways so the three points are power range and data rate so you have to give up one thing in terms of NFC you're giving up on range right it's very short range in terms of Wi-Fi you're giving up on power because you need high power so you so in terms of bandwidth you have to consider what kind of bandwidth you need for mobile it's high bandwidth for sensor networks maybe a low bandwidth will do changes in link capacity that might be due to weather conditions or interference or noise as well so what is your network exposed to is your network exposed to weather conditions is your network just in the indoor very safe condition next is asymmetrical link I believe this means at the upward or the downward link like whether it can be different or it has to be same the upward and the download maybe it's not even downloading anything you know maybe but just like in terms of measuring say pollution of the urban city maybe it just needs an upward link and nothing to download at all next in terms of a network deployment mobility like I said every node is it moving like the ship or the car is it static it can be a hybrid as well some can be moving some can be kind of fixed and this is where the edge of the network can come in handy where it can be a bit of hybrid the wireless technology you have to consider the type of antenna you're using so when it's only directional you cannot have a high throughput but when you have a very directional antenna you have a better throughput so this is something to consider and finally network infrastructure are you gonna have like hands-off and let them manage it themselves kind of autonomous way or are you gonna do some like a single node that kind of manages some of them and that's something you have to decide I thought this was pretty interesting so this paper kind of segregated wireless mesh network into three types one is the man it that I talked about the mobile ad hoc network each nodes are moving the wireless sensor network and wireless mesh network so let's look at the topology remember the topology that I said so and it also compared with wired network so let's say your laptop connected to the ethernet cable will be a wired network so in this case it's a totally static but obviously man it will be a mobile and the other two can be static traffic usually for man it's it's between one node to another node so you can assume that each nodes are the same but for wireless sensor network it will usually be to a sync where kind of the data is kind of gathered and then you go and put it on your cloud-based service then there is also link capacity link capacity can be varying in terms of man it and wireless sensor network like I said it can be exposed to weather interference or noise so once again it can be confusing with the short forms but these are the general short forms to consider in terms of wireless mesh network and then we also have a how do we kind of evaluate you know let's say for Wi-Fi if you want to know if it's a good Wi-Fi network what do you do you do a trace router is it yeah and you calculate the time right the transmission time for one ping so here are some of the criteria to consider so do you want the routing protocol so this means that the data how does the data go transmission is it proactive which means you kind of say from beforehand hey this sensor node every hour or every three hours will just measure the temperature and put it to the access point or it can be reactive as well like you know when you hear a noise and then you put out the sensor value or it can be a hybrid of both so this is something you have to decide for your mesh network what about the network organization if it is flat this means that every node has exactly the same capability and if it is hierarchical it is usually a cluster base that means there will be a single node that will kind of control the others and maybe the rest will kind of communicate with that node how about location awareness they may or may not use localization system so you can think of a sensor node with a GPS location or what you can do you can already configure that node saying that hey you are in this position in hackerspace and you do not have any localization systems embedded in that node so you can be both so these are some things to consider and lastly mobility as well like I said what are the different performance matrix so it's interesting what is the hop count so because you know it's a mesh network hop count so obviously the minimum number of counts it takes is the better next is the transmission count and transmission time if it is if you're not counting the hop and the next one is definitely energy consumption because and it's a challenge in wireless sensor network and mobile ad hoc network because they are moving and they are not connected to powered lines and of course path availability how do you choose the next node to hop on to how do you know it is available and that's the percentage that you count for path availability alright this one was okay this paper basically listed all the different types of routing protocols I think I didn't even oh no I should fill this up okay I'll go home and fill it up but anyway like for Manit there were like 15 different types I think we probably studied all of them I'm just quoting them but here's a summary I have from the paper so from Manit for mobile ad hoc network is a proactive or a reactive routing protocol preferred it is more of a proactive because you know you're moving the nodes are always moving and depending on the position and some events you want to trigger something so that's definitely a proactive is it a flat hierarchy or or hierarchical every node can have the same function a different function it can be both is it location aware yes or no the metrics in this case has to be hops or distance base because once again it's a mesh network and you need to know where to hop and then get on mobility is of course there is for wireless sensor network yep it can be proactive or hybrid actually because some of the sensor node might be connected to the wired flat or hierarchical it can be both in this case metrics energy becomes suddenly very very important because let's say you are using BLE or Zigbee in this in this case to power off and it's me it's powered by battery mobility can once again be yes or no and similar the lastly for wireless mesh network the metrics that becomes important is the transmission time what bandwidth bandwidth yes that's right all right so um lastly how do you design a wireless mesh network routing protocol I think when this paper was written just like my prof the entire research committee was trying to figure out mesh network routing protocols because traditionally it was the start apology like where to hop on to the next note that was like the research challenge so one of them was if there is low mobility use the performance metrics the transmission time just like our Wi-Fi I guess like if it is like not really moving around you can take the time as a performance metrics but if it is moving around you can take the number of hops as the performance metrics and of course lower the better in terms of hardware technology you can consider antenna design and the network connectivity like I said to consider which type of wireless communication to use and based on that you have the base station and access points different and a routing protocol will it be proactive for fixed or will it be hybrid in fact a hybrid seems more and more the approach it seems a bit of both and a link path or that's that was still unclear the paper said so basically the conclusion is that you need a new routing protocol for wireless mesh network it cannot be the same as the start of ology and this is because you need connectivity and everywhere and I think ten years later we can see that there are so many different types of wireless mesh networks being developed you also want a versatility ease of deployment low cost and a network coverage and this is some more reading that I did like Zigbee I just I was just curious like Zigbee I know is a mesh network as well it seems to do an enhanced hierarchical routing protocol to make sure that you know that every of your data is kind of going to the proper place I also did a little bit reading up on Z-Wave routing I think Z-Wave is totally proprietary and yeah they said like devices can communicate to one another by using intermediate nodes to actively route around and circumvent obstacles or radio deadpoints so you can see that the concepts discussed in the paper were put on actively and some of them were like hybrid and like cluster-based protocols that's all I have actually so yeah I would love some questions and discussions maybe like I know so there are some researchers in the audience who did this who's in here anything to say okay actually I was working on like wireless since the networks and stuff for about six or more years okay I actually stopped doing research on networking because I felt that the Singapore community at least in terms of the academic institutions do not seem to appreciate that much about wireless networking because you can't really see it right so yeah so when you look at when you try to kind of sell your work most people outside the research community don't really appreciate like this because they can't see it to then it's just like you know what is the end product you know do you get the data do you get the type of kind of stuff they have but to them to them routing is just like can you just give me the data right but it's something that I really love okay but it will be something similar as well I guess different types of routing protocols and how you decide where to implement what I think most people as in commercial systems at least the ones that I've worked with recently ZigBee and ZigBee and Z-Wave how about in low power when that will be low power when as well okay so why why they're in that work is also still ZigBee yeah right yes yes when you were comparing where you were giving a summary of yes things for WN yes there is further ahead yeah that one why do performance metrics very based on mobility rather why do they be between those two kinds of performance metrics I'm not sure about the links between mobility and, you know, choosing. Because of a high mobility, how do you get expected transmission time? Yeah. Eximated transmission time, okay, yeah. So, you use hops in that case, yeah. And why, and because it's a better metric, that's why you use it when you have loops. Yes, because I think in both the cases, the performance we are looking at is getting the data to the access points. The path of division, you said, is very clear, but there's still some people working on this. Yeah, I think so. Or rather, 10 years ago, maybe, I don't know, Hussien, do you know anything about it, like link and path optimization? Can I ask a question? Yes. What would be the relationship between like current, nowadays we are using TCP and UDP? TCP and UDP is in a different OSI network layer. So, if you Google for OSI, right, Hussien, am I right? No, the expert is here. See, we're spoiling you, right? You shouldn't answer anything until she tries. Okay, don't answer. Let her try first and then... Okay, if I'm wrong, you correct me. So, basically in networking, right, there's this something, oh, okay, you know, what I always do in images, exactly. So, if you have, so there's, okay, I think this is a good one. Yeah, so TCP is basically here. So, if you view image. So, this is something I did. Yeah, physical layer will be something like Ethernet, Bluetooth, 802.11. And all the way up is the application layer which is like HTTP and FTP and something like that. TCP is in the transport layer of the OSI model. And what we are talking about in a lot of times is probably in the physical layer, the data link layer. Okay, yeah. So, because like PRPN, our right, those PPP are currently moving over TCP or JDP or... Yes, so they are at the higher level of networking. Yeah. It's nice to see midi. Midi? Yeah. Yeah. Always in midi. Always in midi. Actually, there are things false under the network layer. Yeah, network layer. Yeah, the routing protocol false. Yeah, but either way, TCP and the networking layer are the different layers. Yeah. Right. So, for wireless mesh networking or wireless sensor network protocols, usually they don't really use the... Oh, it's... Oh, okay. Okay. Common transport layer. So, they use the networking layer. They just send direct frames. Yeah. Probably you can consider it as HTTP. How do they transmit like data in terms of what kind of package? Because there is... It's what I'm used to here. Probably because... I guess the equivalent to that in like the normal world would be if you hooked up a bunch of things to a switch, an Ethernet switch, and instead of trying to send UDP or TCP traffic, you just send raw Ethernet frames. Okay. So, you know your MAC address. You know the MAC address of who you're talking to. You just put data in the frame and you send it out. Put a sender and receive and load the format. Yeah. Then the raw data protocol does not care. Yeah. Yeah. Thanks. I think it's a layer of abstraction that helps segregate it. Yeah. The more accurate way of saying it is you would actually be sending just IP packets. And you don't really have any layer on top of that except for what you design. Okay. Nice. What makes all these layers so special? It's basically abstraction, I would say. What makes this so interesting is that from a ham radius perspective, I could easily take this... I think ham radius is a physical layer, right? Yes, it is. But the whole idea is that I could easily take this. So far, this is the physical aspect of everything. A heavy topic. Encode the thing as sound. All your protocol things are your hardware. The data link, everything. Since this is more of the software routing side, you can even add of course the bandwidth and all that will be limited based on, for example, your radio that you use. But I would say that you can easily do all sorts of mesh network that you don't really need to restrict yourself to the hardware. As long as the equipment, how they do the data link layers, they can agree on that same... Yeah, yeah, absolutely. That's the whole point of it, I think, to have the abstraction layers. Yeah, that's the whole point. Yeah, absolutely. I would say those layers... Yes. In fact, the other... I have been researching about this so many times. Like, when I research about LP-WAN, there is this low power... Oh, God, where is this? I can't find it. But... Yeah, this one. Look at this view image. Look at this. So this is also using the OSI layer that I said. And this is Sigfox and LoraWAN. And this is one by TIN-XM. But you can also see that this is the physical... They are basically defining the physical layer. And on the top is... You can just add on your own stuff. So one of the... The physical layer is because of the broadcasting band that are... Everything. Yes. You are forbidden. But from the transport all the way to vacation, you are free to do whatever you want. Yes, exactly. So they will not define because they want to keep... In fact, most of them will just layer IPv6 on top of it. In fact, Sigfox Low-Pan is the IPv6 over... Low-area network. Yeah. Sigfox Low-Pan. That's why the name Sigfox Low-Pan. Everything uses 802.154. Yeah, exactly. So that protocol is very important. 802.15.4. And whenever they are talking about low power-WAN, you can go for that and safely. So now you know if Singapore announces the IoT network. You know it's a low-power wide-area network. Really? When? I'm saying, when they... Is that a hint? No, no, no, no. No, no, no. I mean, I'm sure Singapore... Is it a pre-announcement? No, no, no. You know. No, I have nothing. So another interesting one is... Remember when I think you brought him here and we went to a seminar by Nordic semiconductors. They make Bluetooth chips. Yes. So they have an IPv6 over Bluetooth. So against everything. IPv6, yeah. Protocols are the same as what you hear normally, but lower level, you can swap out for different things. So... Yeah. Thanks. So CIFOX, the TI is a big company. What is LORA? LORA-WAN is another... I hear it's an open protocol as compared to the rest. LORA-WAN Alliance is like another separate... IoT has so many alliances, like different alliances. So they are also trying to do their own. ZigBee is trying to do their own. Same concept as CIFOX. Yeah. Same concept. Yeah. Yeah, it's more open. They all define the physical layer and a little bit of the data layer. And then the rest are up to you to define. And today, there's also an LTE advanced spec for this sort of thing. Yeah, there is. Exactly. So, yeah, just note to that when they talk about wide area for devices, it is talking about low bandwidth and wide coverage, unlike our mobile phone, which is high bandwidth. But maybe LTE advanced is not the best example, because if you look at the number of protocols that are used in LTE, it's way more than seven layers. Of course. Yeah, yeah, of course. I mean, this is... Simplifying. If you switch your phone, you're already just registered in the tool queue. I think there's three different servers, because you register at different layers. Yeah. Yeah. Yeah, I think that's about it. This paper, like I said, is not technical. You can read through it and, yeah, just understand. What's the impact factor of the job? Impact factor? That wireless mesh networks needs a different kind of routing protocol rather than the conventional Wi-Fi or the Ethernet. Impact factor of the job? Impact factor. Yeah. Oh, it's like a tier one, tier two. What? I have no idea what's that. Academy. Academy. I have no idea what's that. This is to be a rule. This is to be a rule. The papers we love. It should not be asked and it should not be answered. This is why engineers and researchers come together in this meetup. I can read the... Which tier is this reach I prefer? It's actually because I see the format. This one is Springer. Springer. The attribute is... I have no idea. This is tier one? Okay. I can put a tier one. Springer is... And probably when they publish something about LTE, they jump for one month and... Of course. I have no idea. Yeah. Right. Yeah. That's it. Thank you. What happened to... Your joke is the end of the series. Sorry. I'm sorry. I'm sorry. I have no idea.