Fog Computing-II

Summary

In this lecture, the focus is on the evolution from cloud computing to fog computing, emphasizing the need for processing data at the edge of a network rather than relying solely on cloud computing. This approach addresses bandwidth limitations, real-time application requirements, and localized phenomena. The discussion covers various applications, including connected vehicles and smart grids, and highlights the challenges of resource management and security in fog computing. The session also explores the symbiotic relationship between cloud and fog computing, emphasizing the importance of orchestration and synchronization between them.

Highlights

• Fog computing reduces latency by processing data closer to the source, increasing speed for real-time applications. ⏱️
• Real-time applications benefit significantly from fog computing, such as in accident management where instant data processing is essential. 🚗
• The local processing power of fog nodes helps in minimizing the bandwidth burden on cloud computing. 🌐
• Fog computing is integral for applications in smart cities, connected vehicles, and smart grids for local data processing. 🌆
• Orchestration between cloud and fog computing ensures that each platform works to its strengths, enhancing overall system efficiency. 🌩️

Key Takeaways

• Fog computing operates closer to the edge of the network, offering quicker data processing and less latency than traditional cloud computing. 🚦
• It's especially beneficial for real-time applications like traffic management and emergency response which require immediate data handling. 🚨
• Fog computing can help reduce bandwidth usage by processing and aggregating data locally before sending to the cloud. 📉
• Resource management and synchronization between cloud and fog is crucial to ensure efficient functioning of systems. 🔄
• Security remains a key concern in fog computing, requiring careful measures due to its distributed nature. 🔐

Overview

In this talk, we delve into the realm of fog computing and its distinction from traditional cloud computing. This model emphasizes processing at a lower level or edge of the network, preferred for its ability to handle data with reduced latency and enhanced speed, especially vital for real-time applications. The concept of 'fogging' addresses many technological challenges faced by cloud computing alone, by making localized processing a reality.

Fog computing isn't about replacing the cloud but complementing it. By offloading certain tasks to the edge, fog computing reduces overall system load, optimizes bandwidth usage, and enhances the potential for real-time data processing. This is crucial for applications needing immediate feedback such as smart grids, traffic signals, and emergency systems. Connected vehicles are a quintessential example where fog computing really shines by delivering local insights instantly.

The challenges of fog computing are manifold, ranging from resource management to ensuring robust security measures. Since fog devices tend to be less resource-rich than cloud servers, strategic orchestration and synchronization are necessary. Security is highlighted as a significant concern due to the distributed nature of fog networks, but with careful planning, these issues can be mitigated. The session closes on highlighting the cooperative relationship between cloud and fog computing, as their combined strength allows for a holistic approach to modern data management challenges.

Chapters

• 00:00 - 00:30: Introduction to Fog Computing In this chapter, the discussion continues from previous topics on cloud computing, transitioning to an exploration of fog computing.
• 00:30 - 01:00: Need for Fog Computing The chapter 'Need for Fog Computing' discusses the growing necessity for fog computing in various applications. Instead of sending all data and services to the cloud, fog computing allows processing to occur at a lower level, closer to the data source. This approach is beneficial for addressing bandwidth limitations and reducing bandwidth overload, offering a more efficient data processing solution.
• 01:00 - 01:30: Challenges of Cloud Computing The chapter titled 'Challenges of Cloud Computing' discusses the latency issues related to real-time applications in cloud computing. It highlights the importance of edge computing in handling such challenges by pushing certain functionalities closer to the edge of the network, rather than relying solely on central cloud services. This approach minimizes the time taken for processing and feedback, thereby improving efficiency in real-time application scenarios.
• 01:30 - 02:00: Characteristics of Fog Computing Fog computing addresses the issue of not needing to push all data to the cloud, especially in localized applications like connected vehicles and traffic light management, where it interacts with nearby objects.
• 02:00 - 03:00: Comparison of Cloud and Fog Computing In the chapter titled 'Comparison of Cloud and Fog Computing,' the discussion focuses on the handling of vast amounts of data generated by the proliferation of sensors and IoT devices. The traditional approach of sending raw data to the cloud for processing is contrasted with pre-processing or aggregating the data nearby, which is a characteristic of fog computing. This approach can reduce the need to push all data to the cloud, easing the data management process.
• 03:00 - 04:00: Use Cases of Fog Computing The chapter titled "Use Cases of Fog Computing" discusses the concept of fog computing, which involves moving data services and applications closer to the user's location rather than relying solely on the cloud. It is sometimes referred to as edge computing or a distributed phenomenon. The discussion highlights the need to address certain characteristics of the cloud through fog computing to enhance service delivery.
• 04:00 - 05:00: Fog Computing in IoT and Smart Grids The chapter discusses the concept of fog computing within the context of IoT and smart grids. It highlights that similar to cloud computing, fog computing must offer scalability and a pay-as-you-go model to effectively support infrastructure needs. These features are essential for seamlessly integrating cloud benefits into fog systems. The text implies a necessity for fog computing to align with the cloud's initial motivations for providing flexible and efficient service infrastructures.
• 05:00 - 06:00: Challenges in Resource Management The chapter titled 'Challenges in Resource Management' discusses the strategic allocation of resources in a network. It highlights that not all applications should be handled at the network edge. Instead, some tasks might need to be distributed between the cloud and the network edge or intermediate points. This distribution approach poses various challenges, which the chapter will explore further.
• 06:00 - 07:00: Security Challenges in Fog Computing The chapter discusses security challenges in fog computing. Fog devices, unlike cloud devices, are not resource-rich and often consist of intermediate routers or sensing devices. These devices serve as a bridge in sensor deployments, acting as a sync node. In contrast, cloud devices have high resources available at the backend. The chapter also delves into the importance of resource management in this context.
• 07:00 - 07:30: Conclusion In the conclusion, the discussion centers around the significant success of a certain phenomenon, specifically focusing on fog computing. The chapter highlights how cloud computing has effectively facilitated the realization of the potential inherent in IoT devices. This has been achieved by offering scalable resource provisioning and delivering data intelligence derived from extensive datasets.

Fog Computing-II Transcription

• 00:00 - 00:30 hello ah so we will continue our discussion on cloud computing ah rather we will continue our discussion on ah fog computing so what we have seen ah in the in our previous ah
• 00:30 - 01:00 lecture or previous discussion on fog that ah there is a need of ah of of several application to to instead of pushing all the data services and applications to the to the cloud instead whether you can do at a much lower level right there is there is some of the requirements are ah ah due to the bandwidth ah limitations like or ah reducing the bandwidth overload
• 01:00 - 01:30 and in ah some of the things are real time applications right you need to do some real time applications which ah instead of pushing everything to the cloud getting the feedback process etcetera to end of the edge of the network may take ah much time right so in order to handle this ah we need to need to push or there is a requirement of pushing some of this functionalities to the edge of the network or at intermediate level right
• 01:30 - 02:00 and also we have seen that all not all cases we require everything to be pushed into cloud right like ah specially applications like connected vehicles or ah um you are ah that streetlight or traffic light management where ah it is more localized the phenomena is localized right so it is it is more more deal with the ah objects which are in nearby in nearby spaces
• 02:00 - 02:30 so it is it is no there is ah not much requirement pushing all the data to the things again ah a with the with the huge prolification of sensors and for that matter iots ah there is a huge volume of data generated ah where instead of sending the raw data to the cloud for processing there we can do a pre processing or what we say ah some sort of a aggregation
• 02:30 - 03:00 of the information and push it to the cloud for further processing so ah overall in looking all those things there is a need to bring this ah data services application little down from the cloud what we say fogging or fog computing in some in some cases also people say as a ah edge computing or having a distributed phenomena of the things there are few characteristics what we have looked into of cloud which need to be served
• 03:00 - 03:30 here also like scalability ah infinite ah scaling or scaling pay as you go model or metered services and ah that like ah making the infrastructure free type of situation of the things and several other characteristics or cloud provide what was the median motivation of moving towards cloud need to be supported at by the fog also right so that is that is
• 03:30 - 04:00 the need of the things and is not like that that all all the a given a particular application a everything should be put on the edge of the things there may be ah we can do a partially at the at the cloud end and partially at the at intermediate or edge of the network right so this will bring different sort of challenges what we will try to look at ah in ah todays
• 04:00 - 04:30 talk that what are the different type of things what are the fog fog devices are not so ah resource reach right like ah they are ah devices like intermediate routers or at time the sensing devices or the sync of that particular ah sensor deployment sync node of the versus deployment which is not actually so cool at the cloud not no whereas as this was cool at the at the backend high resource cloud so there is a resource management comes in
• 04:30 - 05:00 a big way ah of success of this type of a phenomena right so we will talk about fog computing ah so what we what we see that cloud computing has been able to help in realizing ah the potential of iot devices or iots by providing scalable scalability resource provisioning as well as providing data intelligence from the large
• 05:00 - 05:30 amount of data so one is the scalability ah resource provisioning and the other is it it can basically do some sort of a knowledge mining from the data right making data ah to ah from data to knowledge transformation sort of things right that is at the backend we are machine learning another type of ah algorithms which can run but the cloud has few limitations like specifically in the context of real time latency right
• 05:30 - 06:00 response required in seconds or milliseconds or microseconds sensitive applications likes in case of its a application ah accident say if if there is a collision of the car and if there is this this cars are intelligent car ah what we say that ah having which are regular ad hoc network or sort of things if we need to push this ah this policy and related information to the cloud get it ah refined and find out the location etcetera in from
• 06:00 - 06:30 the nearby cloud that maybe a by that time there may be few more or many more applicants a accident could have happened so it could have been done in a very localized manner where ah instead of taking it to that that cloud and doing the processing i could have done at the localized manner impossible so this sort of real time applications ah where this accident management or some other type of applications what we see that there
• 06:30 - 07:00 can be a ah possible phenomenon doing that so fog computing has been coined in order to serve real time latency sensitive applications faster right it has been coined to serve real time faster fog computing leverages the local knowledge of the data that is available ah to the fog and draws insights from the data by providing faster applications so first of all is a order to serve real time latency ah or real time
• 07:00 - 07:30 applications or which are time sensitive applications and also it has it leverages the local knowledge of the data more at the local level available to the node and draws insight from the data by ah faster response right so that is that exactly it tries to look at that ah fog sort of environment now this picture ah we have seen in the earlier ah our discussion on fog like we have one in cloud one in these ah sensors and mobile
• 07:30 - 08:00 devices another ah what we say contributing to were contributing to the data intermediate ah devices ah can constitute this fog right so this is ah this may not be specially installed for that right so we are having say routers or ah gateways some systems and other devices which are basically communicating to the which are basically used for intermediate communication
• 08:00 - 08:30 of the sensed data to the cloud right so transmitting they are more as a ah some sort of a working as a store and forward from there why whether we can do ah store process forward and take some all locally right some of the things need to be forwarded at the other end ah for higher level of things or maybe aggregation over a over a larger ah geographical space when different sensors coming from the things nevertheless we can have a localized phenomena
• 08:30 - 09:00 ah what we look at at the ah using the data at the local level so what we see that more computing power more storage is at the end the end so the applications which are more computing intensive can be push to the other end whereas more interactive or more responsive can be at the ah lower end so these are the two side of the things so intermediate we have fog or even in some cases this inter frontend devices can do some
• 09:00 - 09:30 some sort of a sort of calculation and aggregate the data and same to the things so if you look at ah if we try to compare or compare and contrast between cloud and fog [a/as] as we have discussed earlier ah um discuss in earlier lecture that these are not ah competitive means in the sense they are not replacing one by another so what it
• 09:30 - 10:00 is trying to do it is mode of a in a companion mode right some of the things which can be done ah better in ah in cloud and can should be done there and fog and they should do in a proper orchestration of the things a a much requirement of the ah orchestration so if you look a if we take look at requirement of cloud and fog so if we say latency so ah is higher in cloud or lower in fog right so latency is a higher delay jitter definitely
• 10:00 - 10:30 the then say high are usually it is high and it is very low in the fog or low in the fog location of server nodes within internet right whereas at the edge of the local network right so where the server nodes are located in case of a cloud it is in the in the internet right you really dont know that where the services are given you talk about ah amazon or ah um
• 10:30 - 11:00 google or microsoft or ibm or anything so sports or anything so what we do we basically connect through through their ah ah portal or link and we really ah as the individual dont know that where your application is being done it is it is ah not like that that you cannot know you can know but nevertheless they do on their resource management and provisioning
• 11:00 - 11:30 and type of thing so we so that is what we say in the internet whereas ah if you look at that fog type of scenario so it is more on the edge in the local network like if i am doing say aggregation of the temperature sensing and taking a design of this particular room with which is having say tain or sensors then what i am doing is basically locally and i know that the server in this particular room is working for that things or the server which is serving for this particular room is working on the things
• 11:30 - 12:00 right at times that is useful because ah if you see ah some congestion or some ah problem in the things you can address the things at times there are other challenges it may be a security loophole also because you know if the temperature at being sensed by the servers and it is sending to things and if this particular lab or room in housing important other systems then i can basically attack that server and say that ah do some manufacturing
• 12:00 - 12:30 right even the temperature or means environmental sensing is giving some allowed i say that everything is going on fine and at that type of thing so there are downfall and type of things ah required and secondly it needs to be resourceful to cater to the type of applications ah which i am trying to do distance between client and the server usually in case of ah cloud computing is multi hop you are using this standard networking to go to the thing usually in case of a cloud it is a one hop may in
• 12:30 - 13:00 case of a fog it is a one hop right so they are at a one hop distance from the from the client to the server right security ah in case of a cloud computing one one i knew that it is undefined un defined in the sense that as the user and i dont have much control over the things right so it is in that sense it is undefined whereas here
• 13:00 - 13:30 it can be defined like you you have a local things you may have some control over the thing so you use the may have some control over these ah particular devices or the organization can have control over the devices and can try to ensure some security like if i say the traffic light so the of a particular city then the traffic light management in that server of a particular zone is under the traffic ah authority of that city right so they have a control had it been on the totally on the cloud so you dont know that what the data
• 13:30 - 14:00 or applications are doing that is that is based on that ah service provider attack on data ah enrouter right so in the enroute data whether the attack in case of cloud if there are multiple hops so there is a chance of much become ah getting much compromised where in case of a fog if it is a single hop then ah as it is a single hop ah then ah ah the getting compromised things are um less right so you have a little more
• 14:00 - 14:30 control over the thing ah location awareness in case of a cloud computing is minimal whereas edge fog computing is location aware it is primarily what we are doing is location aware type of things so there are other ah things like geographical distribution ah in case of a cloud it is more of a centralized feeling so it is having a
• 14:30 - 15:00 logically centralized things in case of a fog it is distributed right number of server nodes ah in case of cloud are few because the that is that is at the clouding and usually servers are extremely resourceful though number of nodes required for the publications or type of things are very few in case of a fog as it is edge as you go down the hierarchy that the number of ah nodes increases ah much more so they are the very large number of
• 15:00 - 15:30 urban nodes ah support for mobility in case of a cloud is limited ah right if you move from one or mobile application that it need to be switched etcetera to the different ah different intermediate devices now it is carried by say one path once you move to the other path whereas in case of a fog ah some sort of support is there in the mobility because because it takes a local deals and other things right it knows a priori
• 15:30 - 16:00 things are that is where under which control and need to be transfer at a much lower level real time applications ah real time interactions ah do supported by cloud and of course it is supported by fog that is one of the major motivation to move towards ah fog computing ah scenarios type of last mile connectivity is usually disliked right so or ah cable line
• 16:00 - 16:30 in case of a in case of cloud right whereas in case of a fog is usually wireless right that there is no ah hard and fast type of things but these are ah you usual standard ah processes in reality right so there are ah pros and cons ah it may so happen that ah as we are discussing fog so it is little bit ah more supportive towards
• 16:30 - 17:00 the fog why it is there and but nevertheless to both has importance and proper a synchronization orchestration between this fog and cloud ah should make the whole thing a reality so there are a several use cases right or several scenarios where fogs are fog will be very much ah applicable one is that emergency evacuation system ah for any catastrophe or disaster real time
• 17:00 - 17:30 information about currently affected areas or buildings and exit route planning etcetera so if it is a large building with several route paths for exit ah if there is a some catastrophe like fire or earthquake or something then the default paths may be getting locked right or ah may not be ah routable so what you need to do you need to do a based
• 17:30 - 18:00 on that availability of the ah exit at every level or every ah individual rooms etcetera we need to path plan the path so we need to have it need need to be dynamically replant or reroute it and where ah may be some local dsm is much more helpful natural disasters management so real time notification about landslides flash flood to potential affected area so that is one requirement so when we are having natural disaster management so
• 18:00 - 18:30 real time ah notification things and there which are ah sometimes pretty localized for a particular region of interest where the things are going on ah and may be ah useful way if we have a this location aware ah information at a fog level so large sensor deployments generate a lot of data which can be pre processed summarized and then to send to the ah cloud to reduce congestion in the intermediate network so
• 18:30 - 19:00 that is another requirement of fog it may not be natural disasters management or natural ah disasters or hazard but in other sense what we have that ah huge deployment of sensors that can have produce lot of data and it takes if you send the all the raw data to the cloud it takes a lot of ah bandwidth and lead to congestion so which can be reduced by moving pushing aggregated data and of course the internet
• 19:00 - 19:30 of things right ah based on big data applications like connected vehicle smart cities wireless sensor network actuators networks and those and its said the so these are ah these are all different different what we aspects of are the scenarios of internet things which again push lot of data and ah in the whole framework and all every time that ah you are
• 19:30 - 20:00 this may not be important like ah that maybe a local dsm is more important than taking a global dsm and type of things and if we look at the applicability there are few here there are hundred more which can be their so smart ah traffic lighting maybe one application connected vehicles ah smart grids of course sensor network internet
• 20:00 - 20:30 of things and software defined network they provide this backbone of these applications to work on all right so if we look at the connected vehicle deployment displays rich scenario of connectivity car to car ah car to access points and type of things so what we look at when we when we talk about vehicular infrastructure we have ah on the road ah different ah moving ah cars right so we ah ah so these are different vehicle
• 20:30 - 21:00 and of course different ah infrastructures so we have connectivity ah between ah this vehicle to vehicle or what we say v two v or v two i i two i so these are different infrastructure which are hovering around and giving connectivity to these different vehicles
• 21:00 - 21:30 so this sort of things are different type of application one is safety related applications another is say information on infotainment information related application and ah there are other things like alats and other type of things based on the title so those are the informations which will be there and in c ah there may be a ah overall
• 21:30 - 22:00 cloud infrastructure where all could have communicated right and it takes a diesel and sends back information type of things now say there is a clash in this particular ah particular vehicle right if there is a clash then other vehicles which are approaching
• 22:00 - 22:30 this vehicle now get information via this instead it could have been done locally right if i can if i can set up a some sort of a fog around this sort of things then i could have taken a local diesel because the accident here may not be nothing to do with some road going somewhere in some cities etcetera right or even in the same city some other part of the thing now here this this type of connected vehicle
• 22:30 - 23:00 phenomena or what we say ah a concept which is ah coming up ah or it is already they are vanet vehicular ad hoc network right so to make it successful this fog maybe one of the applications right so again smart city lighting as we see that if there is a traffic congestion etcetera that within the thing ah within that particular localized things ah sitting so fog has a number of attributes that make idle platform for connected vehicle in providing
• 23:00 - 23:30 services like infotainment safety traffic support and analytics like geo distribution mobility location awareness low latency heterogeneity and so and so forth so there is a lot of applications are there so like ah what we see that there are different vehicles running at the backend so there are the access point or what we say that infrastructure along or sometimes known at road side unit or rsus so these units are there there are
• 23:30 - 24:00 other traffic lights so based on this congestion etcetera this traffic light can be inclusion like it it gives ah the timing for ah stop and go type of things may vary based on the congestion level so and at the backend we have that cloud which takes larger analytics problems which a which requires larger resources and we require a fog orchestration and network management layer which takes care of this synchronization or orchestration of the fog
• 24:00 - 24:30 and also orchestration with the backend cloud so its a system so what we try to see that this fog take a local diesel whereas large analytics can be pushed to the cloud environment right so this is one of the scenarios there of course if we look at a more of a internet of things so at the lower end we have embedded systems and sensors then multi service ah edge ah which takes the distributed in inclusions where fog can played a [row/role] role and
• 24:30 - 25:00 then we have a core network which is which is used to push the things at the upper cloud right so this is ah a in generic way of looking at the internet of things where we see that a fog layer may help in reduce latency and providing better services and though ah in our country may not be ah highly ah proliferated or it is not in a big
• 25:00 - 25:30 way use but it is going to come is that a smart grid right so every ah home will have a smart meter and based on the ah based on utilization of the things at at at a ah from the from the home level ah to from at the house from unit of the house to as say region level and a largest state level the overall management can be done so i have a smartgrid which not only ah give power to the homes and offices and installations also take a
• 25:30 - 26:00 feedback and takes a call based on the things that how the power utilizations are there it is connected to the power plant and the overall power management across the region across the country or across a larger geographical space can be managed by these sort of things so here also fog plays a important role if i this if you are looking as a all things can be pushed into the cloud and take a call right [how/however] however suppose i consider
• 26:00 - 26:30 iit kharagpur if the homes are having smart meters then i could have taken a local decision that what is the overall utilization of the power and ah type of things and i send a aggregated information to the backend ah cloud right so which takes a more ah takes of this aggregated information and take to analytics that over the over larger time span or over days months etcetera how things varies and take a call that what what to provisioning of electricity
• 26:30 - 27:00 based on the things so this type of things are useful where ah we have that in devices or in ah consumer of electricity then we have a fog infrastructure where micro bleeds and other things are there and then we have a micro station and push it to the larger ah in sort of a cloud environment and as we see that all ah all what we say
• 27:00 - 27:30 ah lucrative or golden side of the form there are few definitely challenges or there are ah good amount of challenges in or to have ah have a realization first of all these devices are not that resource full as cloud or cloud servers etcetera so that can give a ah a what we say support to a minimize things so if it is if it is if such application is there which require much larger thing so you need to divide that
• 27:30 - 28:00 biggest and accordingly right so it it may be among the fog devices some portion on this fog devices and some of the cloud and whenever we do this there is a lot of need of synchronization orchestration of the things right because not only now the data are divided your application or the process is also divided so somewhere this ah aggregation of this data and processes need to be there so that is that is one one of the challenges there are other challenges
• 28:00 - 28:30 of the resource management in the fog itself like suppose ah if one of the fog device is overloaded whether i can migrate this application on the things whether i migrate this on the life like executing things can be migrated to the things so these are serious challenges if we if we need to look at the things so let us see some of the things so fog computing system suffers from issue of proper resource allocation among applications while ensuring end to end latency of the services right so what we want to do
• 28:30 - 29:00 that end to end latency and challenges are face resource management of the fog computing network has to be addressed ah so that the system throughput increases ensuring high availability as well as scalability so the basic phenomenon of the fog and finally as as these are distributed over different geographical space may be at with different authorities then what about the security of this application and data and type of things whether that becomes
• 29:00 - 29:30 a source for things so security aspects we discussed last ah last lecture or last discussion on fog that this is this is a serious challenge so resource management in fog network so has different aspect they utilization of idle fog nodes for better throughput whether there is possible there is some some of the fog nodes and it is basically skewed some nodes are more loaded than others more parallel operations how to generate more parallel operations
• 29:30 - 30:00 handling load balancing meeting the delay requirements of real time applications ah right ah so if you have a real time applications how it can be provisions properly provisioning crash fault tolerant and type of things right so like i can say that if the fog node goes down what will happen to those applications and data and which are running on that fog nodes how to handle those how to migrate those data whether i require a a prior in application that even if goes down the other will take up and all those things require a resource
• 30:00 - 30:30 management and incurs cost and so and so forth more scalable systems so at the scalability is our core of the whole thing right scalability is one of the major aspects of cloud we service fog ah computing so how to ah have better scalable systems so they are ah if we look at little more nitty gritty data may not be available at the executing fog node so the application is there the node may not be there therefore data fetching is needed from the
• 30:30 - 31:00 required sensor or data source so that is a phase cyclist their executing node might become unresponsive due to heavy workload which compromises of the latency there may be a issue choosing a new node in case of a micro service execution ah migration so that the response time gets reduced even if i have a way of migration if i am a micro service is running on a particular node or a smaller or cut down version of the service
• 31:00 - 31:30 or chopped service or a partition service is running if i am if even ah if i am if i am able to migrate that i see that this node is going down what should be where i should migrate how to find a node at a thing so some algorithms would run and type of things and some sort of a managements would come into play due to a unavailability of executing node there is a need to migrate partially processed persistent data to a new node so it is hupcup thing need to be migrated so that is another need ah final result has to
• 31:30 - 32:00 be transferred to the client or actuator within less ah amount of time in order in doing so i should not lose out on the time deploying an application components in different fog nodes ensuring latency requirements of the components multiple applications may collocate in the same fog node right therefore the data of one application may get compromised by the other so it may happen that number of application more than one application in the same data security and integrity of individual application
• 32:00 - 32:30 and resource application has to be ensured right so what we look at the ah multi tenancy problem in the cloud that sort of problem can be there in the fog and fog gives less resource ah so that may the problem may be more escalated and there are several approaches people try to follow like ah executing migration on the nearest node which is available ah or to the thing so the nearest node which is free may not be the ah most suitable but
• 32:30 - 33:00 the available node type of things minimizing carbon footprint or video thing my major objective is that to reduce that energy or carbon footprint emphasis on resource prediction whether i can have a the approaches prediction resource estimation reservation ah advanced reservation as well as pricing of the new iot application so that we can do a a priori estimation of the things there is another services which we say docker as an edge computing platform ah to deploying docker may facilitate fast deployment ah elasticity
• 33:00 - 33:30 good performance so resource there are resource management based on ah fluctuating relinquished probability of the customer research prices etcetera people follow that there are other things like studying the base station association tasks distribution virtual machine placement and so an for and ah formulating a lp ah formulation to optimize the thing applying heuristics
• 33:30 - 34:00 ah algorithm to a ah approach that problem so these are the different type of approaches which people are trying to do and ah you can see these are there are lot of research motivation here right there is a there is a there are lot of ah research going on and those who are interested this is a this is a ah field where ah which can be looked into or you can work on those ah type of aspects this is a ah a ongoing another upcoming area to look
• 34:00 - 34:30 at and ah security issues already we have discussed i am not repeating that so security what we try to emphasize here security is also a major challenge right because low resource you cannot run say resource pool or what we say resource hungry security ah applications what we say which ah or security measures which are resource hungry so you need to be ah need to be appropriately sized to this fog type of fog devices to run on
• 34:30 - 35:00 so security is also a major issue it is not only that the data i can be compromised the fog is devices may be a platform to ah simulate ah to simulate or launch attacks right so because it is ah distributed now things are distributed less control over this centralized cloud and the isp so there is a there is a chance of ah this being exploited so need to be looked into things are there how secured or how robust this fog devices are also a
• 35:00 - 35:30 major challenge people are working on with this we let us stop today so what we discussed that that ah this the importance of fog amalgamation of the fog and cloud that it is is not like that ah a through above throughout one of the technology are things are there rather proper synchronization and orchestration between them is a is the real way to have a successful implementation of this framework thank you