IEEE IoT Webinar: Defining 5G Architecture as the Future Communications Infrastructure of IoT Q&A
Presented by Roberto Minerva, Chair, IEEE IoT Initiative; Telecom Italia Lab
February, 24 2016
The definition of the next generation of mobile network is at an initial stage and there is not a consistent and shared view about its architecture and technical aspects. In this early stage it is important to define the goals and requirements of the future infrastructure. Telecom Operators want to play a key role in this and they are putting forward a very detailed set of requirements capitalizing the efforts on M2M architectural definition and the evolution to 5G.
The definition of the next generation of mobile network will not be inertial (simple technological improvements), instead the goal is to position the 5G network as a powerful enabler for many industries and at the center stage of many future business and technological transformation. The 5G network aims at providing a considerable and diversified capacity and intelligent functions to a large set of classes of applications. This webcast will focus on how the different requirements of IoT could help in defining a slice of the forthcoming architecture of the 5G of mobile systems.
Access this webinar on demand and see below for the Q&A session with Roberto Minerva.
Q: SDN is more vulnerable isn’t it?
A: SDN as an open platform has some risks. I believe that opening up APIs has a lot of benefits for several stakeholders (Operators, Service Providers, and possibly users). There is the need to have a strong security framework in order to ensure that risks are mitigated and manageable. Virtualization is also a mechanism that could be used to segment the environments and provide some levels of security.
Q: If I used Windows and linux via vmware may I called this virtualization?
A: Vmware is a suite that supports virtualization. You can use it for virtualizing a Linux distro within a windows system.
Q: How do we relate QoS for 5G & IoT applications?
A: Many IoT applications will have relevant requirements, however, 5G will be an “all-IP” network. So, best effort will be the paradigm to provide services. Possibly the concept of “slicing” could be used in order to create service related slices (i.e., virtualization of network and edge resources) that can fulfill the requirements of those applications. Then QoS will also depend on local devices and by the service applications and these could be “outside” of the 5G infrastructure.
Q: How are these network infrastructures merged with the IoT, connecting the virtual world with the physical world we live in?
A: 5G will be a very pervasive infrastructure offering very important communications capabilities. IoT applications (and related domains such as automotive, e-health, industrial internet, etc.) can use this widely available infrastructure in order to satisfy their communications needs. I think 5G will be an essential enabler for moving towards the evolution that IoT is promising. In addition, 5G seems essential for allowing the mirroring of physical objects with their logical representation in the cloud. Fast communication between physical and virtual objects can enable what I call the Virtual Continuum, i.e., the software entanglement between physical and logical objects. This entanglement can create new opportunities for services and applications. In particular, it allows for the servitization of physical objects, i.e., the possibility of transforming physical objects into services. Think of a dishwasher, if it is connected, you can “rent” it rather than buying it. You pay per use, and you could also have access to services offered by the dishwasher manufacturer.
Q: What would be an application example of virtualization on the edge side?
A: I think the best example is in the field of autonomous cars and vehicles. Vehicles can organize themselves in order to move as small “trains”. This requires a lot of coordination between the edge devices. Robotics is another interesting field of investigation, as well as disaster recovery where an edge device can self-organize in order to create environments that can support communications and exchange information. I believe that edge computing is still to be fully investigated in order to unleash a well of possibilities. Keep in mind that creating dynamic clouds at the edge means also to bypass current infrastructure offered by big service providers and entering into new business models supported by the cooperation of crowds of people.
Q: If the data is gathered from sensors in IoT and transferred over 5G, how can we guarantee the speed promised?
A: 5G networks will be very capable of reaching speed and bandwidth capabilities. We need to use virtualization wisely in order to fulfill the requirements of the envisaged applications. The constant improvement of performance of general purpose hardware promises to be able to support the envisaged speeds of 5G.
Q: What do you forsee to be the impact of the recently announced "passive wi-fi", using 10,000 less energy?
A: There are many different protocols and solutions competing in the IoT realm, especially in the short-range communications sector. This kind of competition is good because they will try to make it possible for sensors to last longer by consuming less power. Passive Wi-Fi seems to promise a huge saving in energy consumption and, at the same time, allows the usage of a well-known technology. However, I have the feeling that at the edge, many competing technologies will coexist depending on the different features they provide and on specific service requirements.
Q: By edge networking, do you mean something like creating a cluster of edge nodes, which can do cloud computing like analysis of Big Data etc.?
A: Yes, I see edge computing as the possibility of creating dynamic distributed systems capable of supporting communications, processing, storage, and sensing/actuation within smart environments. Data produced within these “confined” environments could be manipulated locally but they could also be stored in larger clouds for backend processing. The importance of edge computing is to make use of relevant local capabilities to directly serve a community of users.
Q: Do you think that a "Social Network" for things/devices could promote altruistic behaviour?
A: Yes, I see a strong relationship between edge computing, social networks, and peer-to-peer networks. In order for p2p networks to be successful there is the need for altruistic behavior (e.g., sharing of data, allowing the usage of communications capabilities and the like). The more altruistic the behavior the more the “edge network” is useful for the community. In social networking studies, there are attempts to determine when an altruistic behavior can be induced on network nodes.
Q: Would every device be a subscriber with a different U-SIM Card?
A: No, I think there will be many devices that will not use USIM. USIM can provide some security levels and be a kind of encryption point and frontend to the public network, a sort of anchor point that other nodes can rely on for long haul communications. Depending on the situation and the services, some devices will have a USIM and others not.
Q: What area would be good for new research in 5G?
A: I think there are a few areas of interesting research: new radio and heterogeneous communications poses a great deal of challenges (cognitive radio could be a further disruption in this context), but also autonomics and self-organization should be considered in order to limit human intervention in managing large networks that will be by definition highly dynamic and unpredictable. In addition, a new control software architecture encompassing the edge and core of the network should be considered. Finally, I think 5G will be a service enabler and more effort should be put on understanding the “service science” and how to design 5G and future networks in order to be capable of supporting new services.
Q: Can you please discuss how cloud computing and fog/edge computing interact and co-exist in the 5G scenario?
A: Edge/fog computing are very similar to what p2p networks are, i.e., networks in which each node can provide or request services to others. I see cloud computing more as an implementation of the client – server paradigm. Services can be implemented in both environments. Certain services are better in p2p (think of Skype or the file download), others are better in the C-S environment. Fog/Edge Computing and cloud computing will be used also in hybrid solutions in order to exploit the capabilities that they can offer. I think it is up to the service designer to figure out how the service should be implemented. The designer should choose the best paradigm supporting the service and also consider processing power, communications costs and the like. But undoubtedly there are so many resources at the edge, that we have the need to make a wise use of them.
Q: What do IPSO, OIC, AllSeen, etc. mean to 5G?
A: They propose specific usage and mechanisms for supporting IoT applications. From my perspective, 5G can satisfy their communication needs. In other terms it will be an enabler for their solutions and deployment.
Q: Doesn't this vision depend on IPv6 everywhere? IPv4 is still dominant? Won't this hold back progress?
A: I hope IPv6 will take over in this domain. However, many services and application will still be possible by using IPv4. My hope is to have IPv6 widely deployed in order to take full advantage of its capabilities.
Q: Can we define IoT as a network of wireless sensor networks? What will be the impact on Quality of Service?
A: I think we need to make a distinction between IoT systems. In small environments (also with many sensors), IoT will be a wireless sensor network with a service center able to provide services. This is strongly related to the notion of a single administrative domain, i.e., a single entity will manage the system. When more than a system will cooperate, and these systems pertain to different administrative domains, then the IoT definition becomes more complex and challenging. I like this definition of Internet of Things (you can find it in this document: http://iot.ieee.org/definition.html) : The Internet of Things (IoT) envisions a self-configuring and adaptive complex system made out of networks of sensors and smart objects whose purpose is to interconnect “all” things, including everyday and industrial objects in such a way as to make them intelligent, programmable, and more capable of interacting with humans by providing useful services.
For QoS see previous questions.
Q: As you mentioned 5G has to deal with low latency, so what are the approaches we can think of achieving low latency in SDN?
A: Edge and fog computing are one of the means for dealing with low latency. Having strong computing capabilities at the edge means that processing can occur locally, data can be stored and cached locally. However, we have also to design the core network in a slim and efficient way in order to decrease the latency in control and data transport.
Q: Will smart antennae and beam-forming be integral to 5G and will the cost be supportable?
A: Definitely yes. You may be interested in these IEEE articles: http://spectrum.ieee.org/telecom/wireless/smart-antennas-could-open-up-new-spectrum-for-5g and http://cms.comsoc.org/SiteGen/Uploads/Public/Docs_TC_5GMWI/Networks_and_Devices_for_the_5G_Era.pdf.
If there is an industrial requirement, probably costs will progressively decrease and become sustainable.
Q: How can the issue of interference from similar network sources using 5G be handled?
A: Interference is for sure an issue in 5G considering also the fact that the access will be very heterogeneous and many more antennas are to be deployed. A good survey of these issues and some possible strategies are presented in the book: Interference Coordination for 5G Cellular Networks by Lu Yang and Wei Zhang. They point on; beamforming techniques, power control, user scheduling, and advance receiver techniques. From my perspective all these techniques should be governed in an autonomic manner from the control structure of the 5G architecture.
Q: It would appear that in the event, although slim probability, of a catastrophic failure occurring it would become nearly impossible to reconstruct the failure in terms of Sequence of Events or single point of failure analysis.
A: If we see the IoT as a complex system, we may need to approach the disaster recovery issues from a different perspective. I think autonomic solutions and self configuration of large IoT systems will be a strong requirement. Also the 5G network due to its heterogeneity (especially at the access) should benefit from these properties.
Q: Will 5G accommodate new technologies coming up in the wireless space? i.e. White space frequency bands?
A: I think one of the goals of 5G is exactly to accommodate new technologies in the wireless space. With respect to Whitespaces, I think it is also a regulation issue. Some authorities are interested in the usage of that spectrum for wireless broadband. Ofcom (UK) seems to be strongly oriented to this (http://www.belfasttelegraph.co.uk/technology/tv-white-space-wireless-broadband-plans-for-new-kind-of-mobile-internet-that-could-pave-way-for-5g-revolution-30987645.html).
Q: Will the legacy software stacks developed for 4G co-exist with 5G?
A: I see a problem in this approach especially in the IMS part. IMS was designed and developed many years ago. Using it in a network that will be deployed in the next decade could be useless. I think 5G should be an opportunity to close the software gap between current telecom approaches and the evolution of software technologies. 5G should jump start new things. Backward compatibility could be supported for a while by using virtualization of old systems.
Q: MIMO tx/tx for iot in 5g space
A: Most likely, yes. There is work going on at the academic level (e.g., Choi, Junil, et al. "Distributed reception with spatial multiplexing: MIMO systems for the Internet of Things." arXiv preprint arXiv:1409.7850 (2014).) and at the industrial level (http://www.taoglas.com/article/taoglas-unveils-first-embedded-4x4-mimo-antenna-for-iot-and-m2m-applications).
Q: What kind of storage system or enhancements are expected in IoT, i.e. how will traditional database systems have to change to accommodate so frequent and such huge data?
A: IoT applications will cover a very broad set of application domains. Most probably there will be the need to use more than one paradigm for dealing with massive data handling. However, Big Data techniques seem to be very important and IoT is seen by many experts as one of the most important applications for big data. From my perspective it would be important to consider and integrate different communication paradigms (e.g., PubSub) within IoT and 5G systems. The creation of Complex Event Processing engines is quite essential for many IoT applications. This should also come with a more appropriate handling of large real-time data streams. I find that the concept of stream computing from IBM (http://www.ibm.com/analytics/us/en/technology/stream-computing/?lnk=ushpv18f2l) fits quite well with the requirements of large IoT applications.
Q: What are the future /promising research / design areas in IoT / Big Data era?
A: I see Complex Event Processing engines, stream computing, and a lot of AI techniques for extracting valuable information from the IoT wealth of data. Imagine the data collection related to the “working” of a city and the effort and techniques needed to extract valuable information about the city itself, or other relevant information.
Q: What in your opinion will be a dominant technology for enabling 5G?
A: I think the core of evolution for 5G will be on radio technologies. Then a lot of effort should go on softwarization of the network in order to make 5G networks programmable, flexible, and secure covering the software gap with systems operated by Web Companies.
Q: If I want to start to learn about IoT where can I get information about it?
A: There are a multitude of opportunities to learn more about IoT on the IEEE IoT web portal http://iot.ieee.org. The IEEE IoT Initiative produces and makes available on-demand topical webinars, a newsletter, conferences and events, educational videos, and interactive platforms such as the very comprehensive living document, Towards a Definition of the Internet of Things, and the IoT Scenarios program that describes IoT services, applications, business models, etc. If you join the IEEE IoT Technical Community, you will receive additional information such as the opportunity to view a monthly selection of IoT-related articles from IEEE Xplore. The mission of the IEEE IoT Initiative is to serve as the gathering place for the global technical community working on the Internet of Things; to provide the platform where professionals learn, share knowledge, and collaborate on this sweeping convergence of technologies, markets, applications, and the Internet, and together change the world. We encourage your participation!