Phil Sheppard, Director of Network Strategy & Architecture, Three UK was the keynote speaker of LTE Voice Summit held in London this month. Its been over a year that Three launched its VoLTE service in the 800MHz band. In fact recently, it has started showing adverts with Maisie Williams (Arya Stark from Game of Thrones) fighting black spots (not spots) with 4G Super-Voice.
As I highlighted in the LTEVoice 2015 summary where China Mobile group vice-president Mr.Liu Aili admitted "VoLTE network deployment is the one of the most difficult project ever, the implementation complexity and workload is unparalleled in history", Three UK's experience wasn't very different. Quoting from ThinkSmallCell summary of the event:
It was a huge project, the scope far exceeding original expectations and affecting almost every part of their operations. They spent 22,245 man days (excluding vendor staff time) – more than 100 man years of effort – mostly involved with running huge numbers of test cases on the network and devices.
There are some other interesting bits from the different summaries that are provided in references below but here are few things I found of interest with regards to Three UK VoLTE deployment:
170 million voice calls minutes have used VoLTE since the launch in Sept 2015
Only devices that can support VoLTE and 800MHz are allowed to camp on 800MHz band. This is to avoid disappointment with CS Fallback
There are plans to roll out VoLTE in other bands too once all niggles are ironed out in the 800MHz band.
The summary offers insight to the future of wireless broadband and how new requirements and technological goals will be achieved. The report updates Release 13 (Rel-13) features that are now completed at 3GPP and were not available at the time of the publication of a detailed 5G Americas report, Mobile Broadband Evolution Towards 5G: 3GPP Release 12 & Release 13 andBeyond in June 2015.
The 3GPP standards have many innovations remaining for LTE to create a foundation for 5G. Rel-12, which was finalized in December 2014, contains a vast array of features for both LTE and HSPA+ that bring greater efficiency for networks and devices, as well as enable new applications and services. Many of the Rel-12 features were extended into Rel-13. Rel-13, functionally frozen in December 2015 and completed in March 2016, continues to build on these technical capabilities while adding many robust new features.
Jim Seymour, Principal Engineer, Mobility CTO Group, Cisco and co-leader of the 5G Americas report explained, “3GPP Release 13 is just a peek behind the curtain for the unveiling of future innovations for LTE that will parallel the technical work at 3GPP on 5G. Both LTE and 5G will work together to form our connected future.”
The numerous features in the Rel-13 standards include the following for LTE-Advanced:
Active Antenna Systems (AAS), including beamforming, Multi-Input Multi-Output (MIMO) and Self-Organizing Network (SON) aspects
Enhanced signaling to support inter-site Coordinated Multi-Point Transmission and Reception (CoMP)
Carrier Aggregation (CA) enhancements to support up to 32 component carriers
Dual Connectivity (DC) enhancements to better support multi-vendor deployments with improved traffic steering
Improvements in Radio Access Network (RAN) sharing
Enhancements to Machine Type Communication (MTC)
Enhanced Proximity Services (ProSe)
Some of the standards work in Rel-13 related to spectrum efficiency include:
Licensed Assisted Access for LTE (LAA) in which LTE can be deployed in unlicensed spectrum
LTE Wireless Local Area Network (WLAN) Aggregation (LWA) where Wi-Fi can now be supported by a radio bearer and aggregated with an LTE radio bearer
Narrowband IoT (NB-IoT) where lower power wider coverage LTE carriers have been designed to support IoT applications
Downlink (DL) Multi-User Superposition Transmission (MUST) which is a new concept for transmitting more than one data layer to multiple users without time, frequency or spatial separation
“The vision for 5G is being clarified in each step of the 3GPP standards. To understand those steps, 5G Americas provides reports on the developments in this succinct, understandable format,” said Vicki Livingston, Head of Communications for the association.
I have been covering the LTE Voice Summit for last couple of years (see here: 2015 & 2014) but this year I wont be around unfortunately. Anyway, I am sure there will be many interesting discussions. From my point of view, the 2 topics that have been widely discussed is roaming and VoWiFi.
One of the criticisms of VoWiFi is that it does not the QoS aspect is missing, which makes VoLTE special. In a recent post, I looked at the QoS in VoWiFi issue. If you haven't seen it, see here.
Coming back to VoLTE roaming, I came across this recent presentation by Orange.
This suggests that S8HR is a bad idea, the focus should be on LBO. For anyone who is not aware of the details of S8HR & LBO, please see my earlier blog post here. What this presentation suggests is to use LBO with no MTR (Mobile Termination Rates) but instead use TAP (Transferred Account Procedures). The presentation is embedded below:
Another approach that is not discussed too much but seems to be the norm at the moment is the use of IP eXchange (IPX). I also came across this other panel discussion on the topic
IPX is already in use for data roaming today and acts as a hub between different operators helping to solve inter-operability issues and mediating between roaming models. It can work out based on the calling and callee party what kind of quality and approach to use.
Here is the summary of the panel discussion:
Hopefully the LTE Voice Summit next week will provide some more insights. I look forward to hearing them.
Last year I posted an excerpt from an article in FT which implied that Quantum technology will play a big role in post-5G world. Earlier this month CW held their annual Technology & Engineering Conference (CW TEC). The topic was "The Quantum Revolution is coming". I have to admit that I knew next to nothing before the conference, however now I hope I know just enough to dabble in quantum technology related discussions.
The main question that I had before the conference was 'when will quantum technology be here?'. While there were different answers, depending on what you think Quantum is, I think the answer I feel comfortable is more like 2030 (just in time for 6G?)
There are already some great write-ups of the conference by others, please see links at the bottom of the post. Here are the presentations from the event:
Turning on WMM (or WME) at access point provides significant protection for voice traffic against competing wireless data traffic
Turning on WMM at the client makes only a small difference where there are a small number of clients on the wireless LAN. This plus the “TCP Unfairness” problem means that it can be omitted.
All Home gateways support WMM but their firmware may need to be altered to prioritise on DSCP rather than layer two
Wireless Multimedia Extensions (WME), also known as Wi-Fi Multimedia (WMM), is a Wi-Fi Alliance interoperability certification, based on the IEEE 802.11e standard. It provides basic Quality of service (QoS) features to IEEE 802.11 networks. WMM prioritizes traffic according to four Access Categories (AC): voice (AC_VO), video (AC_VI), best effort (AC_BE), and background (AC_BK). However, it does not provide guaranteed throughput. It is suitable for well-defined applications that require QoS, such as Voice over IP (VoIP) on Wi-Fi phones (VoWLAN). WMM replaces the Wi-Fi DCF distributed coordination function for CSMA/CA wireless frame transmission with Enhanced Distributed Coordination Function (EDCF). EDCF, according to version 1.1 of the WMM specifications by the Wi-Fi Alliance, defines Access Categories labels AC_VO, AC_VI, AC_BE, and AC_BK for the Enhanced Distributed Channel Access (EDCA) parameters that are used by a WMM-enabled station to control how long it sets its Transmission Opportunity (TXOP), according to the information transmitted by the access point to the station. It is implemented for wireless QoS between RF media.
This blog post describes how the QoS works in case of WMM.
Finally, this slide from Cisco shows how it will all fit together.
You have probably read about the demanding requirements for 5G in many of my blog posts. To meet these demanding requirements a 'next-generation radio' or 'new radio' (NR) will be introduced in time for 5G. We dont know as of yet what air interface, modulation technology, number of antennas, etc. for this NR but this slide above from Qualcomm gives an idea of what technologies will be required for this 5G NR.
The slide above gives a list of design innovations that will be required across diverse services as envisioned by 5G proponents.
It should be mentioned that Rel-10/11/12 version of LTE is referred to as LTE-Advanced and Rel-13/14 is being referred to as LTE-A Pro. Rel-15 will probably have a new name but in various discussions its being referred to as eLTE.
When first phase of 5G arrives in Rel-15, eLTE would be used for access network and EPC will still be used for core network. 5G will use NR and eventually get a new core network, probably in time for phase 2. This is often referred to as next generation core network (NGCN).
The slides below from Deutsche Telekom show their vision of how operators should migrate from eLTE to 5G.
The slides below from AT&T show their vision of LTE to 5G migration.
Eiko Seidel posted the following in 3GPP 5G standards group (i recommend you join if you want to follow technical discussions)
Summary RAN1#86 on New Radio (5G) Gothenburg, Sweden At this meeting RAN1 delegates presented and discussed numerous evaluation results mainly in the areas of waveforms and channel coding. Nonetheless RAN1 was not yet prepared to take many technical decisions. Most agreements are still rather general. First NR terminology has been defined. For describing time structures mini-slots have been introduced: a mini-slot is the smallest possible scheduling unit and smaller than a slot or a subframe. Discussions on waveforms favored filtered and windowed OFDM. Channel coding discussions were in favor of LDPC and Turbo codes. But no decisions have been made yet. Not having taken many decisions at this meeting, RAN1 now is behind its schedule for New Radio. Hopefully the lag can be made up at two additional NR specific ad hoc meetings that have been scheduled for January and June 2017. (thanks to my colleague and friend Dr. Frank Kowalewski for writing this short summary!)
Yet another post from Eiko on 3GPP RAN 3 on related topic.
The RAN3 schedule is that in February 2017 recommendations can be made for a protocol architecture. In the meeting arguments came up by some parties that the work plan is mainly addressing U-Plane architecture and that split of C- and U-plane is not considered sufficiently. The background is that the first step will be dual connectivity with LTE using LTE RRC as control plane and some companies would like to concentrate on this initially. It looks like that a prioritization of features might happen in November timeframe. Beside UP and CP split, also the functional split between the central RAN node and the distributed RAN node is taking place for the cloud RAN fronthaul interface. Besides this, also discussion on the fronthaul interface takes place and it will be interesting to see if RAN3 will take the initiative to standardize a CPRI like interface for 5G. Basically on each of the three interfaces controversial discussion is ongoing. Yet another basic question is, what is actually considered as a “New 5G RAN”? Is this term limited to a 5G eNB connected to the NG core? Or can it also be also an eLTE eNB with Dual Connectivity to 5G? Must this eLTE eNB be connected to the 5G core or is it already a 5G RAN when connected to the EPC?
Finally, a slide from Qualcomm on 5G NR standardization & launch.
I have written about Fronthaul as part of C-RAN in this blog as well as in the Small Cells blog. I am also critical of the C-RAN concept now that the Baseband Units (BBU) have become small enough to go on the cell cite. I have expressed this view openly as can be seen in my tweet below.
While I am critical of the C-RAN approach, there are many vendors and engineers & architects within these vendors who are for or against this technology. I am going to leave the benefits and drawbacks of C-RAN in light of new developments (think Moore's law) for some other day.
The above picture from my earlier post explains the concept of Fronthaul and Backhaul for anyone who may not be aware. As data speeds keep on increasing with 4G, 4.5G, 4.9G, 5G, etc. it makes much more sense to use Fiber for Fronthaul. Dark fiber would be a far better choice than a lit one.
One thing that concerned me was what happens in case of MIMO or massive MIMO in 5G. Would we need multiple Fronthaul/Fibre or just a single one would do. After having some discussions with industry colleagues, looks like a single fiber is enough.
This picture above from an NTT presentation illustrates how WDM (Wavelength Division Multiplexing) can be used to send different light wavelengths over a single fiber thereby avoiding the need to have multiple of these fibers in the fronthaul.
There are 2 different projects ongoing to define 5G Fronthaul & Backhaul.
The first of these is 5G Crosshaul. Their website says:
The 5G-Crosshaul project aims at developing a 5G integrated backhaul and fronthaul transport network enabling a flexible and software-defined reconfiguration of all networking elements in a multi-tenant and service-oriented unified management environment. The 5G-Crosshaul transport network envisioned will consist of high-capacity switches and heterogeneous transmission links (e.g., fibre or wireless optics, high-capacity copper, mmWave) interconnecting Remote Radio Heads, 5GPoAs (e.g., macro and small cells), cloud-processing units (mini data centres), and points-of-presence of the core networks of one or multiple service providers. This transport network will flexibly interconnect distributed 5G radio access and core network functions, hosted on in-network cloud nodes, through the implementation of: (i) a control infrastructure using a unified, abstract network model for control plane integration (Crosshaul Control Infrastructure, XCI); (ii) a unified data plane encompassing innovative high-capacity transmission technologies and novel deterministic-latency switch architectures (Crosshaul Packet Forwarding Element, XFE).
5G-XHaul proposes a converged optical and wireless network solution able to flexibly connect Small Cells to the core network. Exploiting user mobility, our solution allows the dynamic allocation of network resources to predicted and actual hotspots. To support these novel concepts, we will develop:
Dynamically programmable, high capacity, low latency, point-to-multipoint mm-Wave transceivers, cooperating with Sub-6 GHz systems;
A Time Shared Optical Network offering elastic and fine granular bandwidth allocation, cooperating with advanced passive optical networks;
A software-defined cognitive control plane, able to forecast traffic demand in time and space, and the ability to reconfigure network components.
The well balanced 5G-XHaul consortium of industrial and research partners with unique expertise and skills across the constituent domains of communication systems and networks will create impact through:
Developing novel converged optical/wireless architectures and network management algorithms for mobile scenarios;
Introduce advanced mm-Wave and optical transceivers and control functions;
Support the development of international standards through technical and technoeconomic contributions.
The differences are summarised in the document below:
The above picture is a summary of the spectrum that was agreed to be studied for IMT-2020 (5G). You can read more about that here. I have often seen discussions around how much spectrum would be needed by each operator in total. While its a complex question, we cannot be sure unless 5G is defined completely. There have been some discussions about the requirements which I am listing below. More informed readers please feel free to add your views as comments.
Real Wireless has done some demand analysis on how much spectrum is required for 5G. A report by them for European Commission is due to be published sometime soon. As can be seen in the slide above, one of the use cases is about multi gigabit motorway. If the operators deploy 5G the way they have deployed 4G then 56 GHz of spectrum would be required. If they move to a 100% shared approach where all operators act as MVNO and there is another entity that deploys all infrastcture, including spectrum then the spectrum requirement will go down to 14 GHz.
This is in addition to all the other spectrum for 2G, 3G & 4G that the operator already holds. I have embedded the presentation below and it can be downloaded from here:
The UK Spectrum Policy Forum (UKSPF) recently held a workshop on Frequency bands for 5G, the presentations for which are available to download on the link I provided.
Its going to be a huge challenge to estimate what applications will require how much amount of spectrum and what would be the priority as compared to other applications. mmMagic is one such group looking at spectrum requirements, use cases, new concepts, etc. They have estimated that around 3.1GHz would be required by each operator for 99% reliability. This seems more reasonable. It would be interesting to see how much would operators be willing to spend for such a quantity of spectrum.
5G is often seen as a panacea for everything that is imperfect in mobile technology. Any issues with coverage, capacity, connectivity and speed are all expected to be solved with the arrival of 5G. While I don’t think we will be able to solve all the issues on the table, 5G will hopefully resolve quite a few of them.
Back in June I did an interview with the organizers of 5G World Series where I expressed my views for the questions that were posed to me. You can see this interview below.
Now that I have had time to think about the questions, here are a bit more detailed thoughts. As always, feedback, comments & suggestions welcome
Q: What will network architecture look like in the 5G era?
I have long argued that 5G will not be a single technology but a combination of multiple old and new technologies. You will often find various terms like Multi-stream Aggregation (MSA), Opportunistic Aggregation and Multi-connectivity being used to explain this. Not only will 2G, 3G and 4G have a role to play, Wi-Fi and other unlicensed technologies would be a part of 5G too.
I have had many discussions on this topic with respected analysts and many of them agree.
This #5G chart sums it up. Start from where we are, go in 4 different directions at once, give it all the same name https://t.co/ETldL8Sr8N
One of the approaches being proposed for the initial version of 5G is the non-standalone version of 5G which will use LTE as the control plane anchor and new 5G radio for user plane. Not only will this be easier to deploy along with the existing LTE network, it would be faster and hopefully less costly.
Q: To what extent is 5G dependent on virtualization?
Networks and Network Functions are progressively being virtualized, independently of 5G. Having said that, virtualization will play a big role in achieving the 5G architecture. Mobile operators can’t be expected to keep paying for proprietary hardware; virtualization would help with cost reduction and quick deployments.
Network slicing for instance will help partition the network for different requirements, on the fly depending on what is going on at any particular time.
Q: What is your view on the interplay between standards and open-source developments?
Standards enable cost reduction by achieving economy of scale whereas open-source development enable innovation and quick deployment. They are both needed and they will willingly or unwillingly co-exist.
Q: What do you see as the 3 greatest technical uncertainties or challenges on route to 5G?
While there are many known and unknown challenges with 5G, some obvious ones that we can see are:
Spectrum identification and harmonization.
Getting to the right architecture which is backward compatible and future proof, without making it too complex
SON – Once you have everything in place you have to make many different parts of the network work together with different kinds of loads and traffic. SON will play a crucial role here.
Q: What would 5G actually mean for consumers, business and IoT? / What will 5G allow me to do that I can’t right now with 4G?
There are a lot of interesting use cases being discussed like remote operations and remote controlled cars but most of them do not represent the general consumers and some of them are just gimmicks.
I really like the NGMN whitepaper that laid out some simple use cases.
If done properly, 5G will allow:
Simplification of the network resulting in low latency – this means that your content will load faster and the delay between requests and responses are small.
Reasonable speed broadband everywhere - This will also depend on the operators’ rollouts plan but different technologies in 5G network would (should) enable a good speed reliable broadband not just in the middle of the cell but also on the edges. In fact, the concept of edges should be looked at in 5G and a solution to avoid data rates falling off should be found.
Connectivity on the move – Whether we are talking about connectivity in trains/buses or from public safety point of view, it is important to define group connectivity, direct communications, etc.
Q: What will set companies apart in the development of 5G?
The days of vendor lock-ins are over. What will set companies apart is their willingness to be open to working with other companies by having open API’s and interfaces. Operator networks will include solutions from many different vendors. For them to be quick to bring innovative solutions to the market, they need vendors to work together rather than against each other.
Q: There is a lot of talk about the vision for 2020. What do you think the world will look like in terms of connectivity in 2030?
It would be fair to say that by 2030, connectivity would have reached a completely new dimension. One of the big areas of development that is being ignored by mainstream mobile community is the development of satellite communications. There are many low earth orbit (LEO) constellations and high-throughput satellites (HTS) being developed. These LEO and HTS combination can provide high speed connectivity with 4G like latency and high throughputs for planes/ships which cannot be served by ground based mobile technology. Broadband access everywhere will only become a reality with satellite technology complementing mobile technology.
Disclaimer: This blog is maintained in my personal capacity and this post expresses my own personal views, not the views of my employer or anyone else.
Looking at a paper (embedded below) from NTT Docomo technical journal where they talk about Dedicated Core Network (DCN) for handling different traffic type (M2M/IoT for example). Note that this approach is different from NFV based network sliced architecture. For the latter, the network functions should have been virtualized.
There will be some signalling overhead in the core network to handle the new core and reroute the traffic according destined for the new dedicated core. I would still hope that this would be minuscule in the grand scheme of things. Anyway, let me know what you think about the paper below.
