Category: RAN

Handover: The Biggest Asset of Wireless Operators in an IP World

For my upcoming course on LTE services at the University of Oxford in April, I've been giving the voice over IP topic some more thought. Unlike UMTS, LTE is a pure IP based network so it doesn't have an inherent circuit switched voice capability. It's a bit like burning the bridges behind you so you can't go back.

From an operator point of view, potential solutions are the IMS (IP Multimedia Subsystem) or reusing the circuit switched MSC architecture over an IP based channel such as promoted by the VOLGA forum, an approach that I think has a great potential! But what are the benefits of network operator VoIP vs. other purely Internet based alternatives such as Skype or about Grand Central / Google Voice, about which Ajit Jaokar has written recently in conjunction with VOLGA? Well, handovers are!

It's handovers because Internet based Voice over IP services will work well while the network can handover a moving user to another LTE or UMTS cell. But as soon as UMTS and LTE coverage runs out and the system is forced to go to the GSM/GPRS/EDGE network that voice call is history.

Not so with network operator based voice systems. For both IMS and VOLGA, methods are in place to prepare a circuit switched channel in the GSM network before the handover and the mobile device is instructed to use this channel after the handover. In the case of VOLGA, it's a pretty straight forward thing to do. Here, higher layers of the protocol stack will not see a difference between the voice call being transported over an IP data stream over LTE and a circuit switched timeslot in the GSM network.

In my opinion, an invaluable advantage for wireless operators that Internet based voice services will not be able to mimick. And I agree with Ajit, in the future we will see application layer based voice services such as Grand Central and network layer based voice services of wireless operators working together instead of fighting with each other.

Escaping Future Bandwidth Bottlenecks: LTE and HSPA on Several Bands

I think everyone in the industry is pretty clear by now that the amount of data that cellular wireless networks will have to carry in the future is going to rise. In my recent book I’ve taken a closer look at theoretical and practical capacity on the cellular level in chapter 3 and I come to the conclusion that from a spectrum point of view, there is quite a lot of free space left in most parts of the world that will last for quite some time to come.

So while alternative approaches like integrating Wi-Fi and femtos into an overall solution will ultimately bring much more capacity, I think it is quite likely that network operators will over time deploy their cellular networks in ever more bands. In Europe, for example, I think it’s quite likely that operators at some point will have networks deployed on the 900, 1800, 2100 and 2600 MHz band simultaneously.

Quite an interesting challenge to solve for networks and especially for mobile devices as they have to support an ever growing number of frequency bands.  Also, those bands should not also be used in tight cooperation instead of just aside each other. Ideally, the resources in the 900 MHz band could be reserved for in-house coverage as radio waves in this band penetrate walls quite well. But as soon as the network or the device detect that other bands can be received quite well, they should automatically switch over to them to leave more capacity for devices used indoors or under difficult radio conditions.

Switching between different frequencies and radio technologies during a call or a session is already done today but mostly based on deteriorating reception levels. So in the future, when using so many bands, I think this reactive mechanism has to be enhanced into a proactive mechanism and switch-overs need to be timed so that the user does not notice an interruption.

How Can LTE Reduce the Cost Per Bit?

Recently, a question was asked in the LTE forum on LinkedIn how LTE can reduce the cost per bit compared to todays broadband wireless systems such as HSPA. I found it quite interesting that a lot of people immediately jumped at the greater spectral efficiency as the means to reduce the overall cost. But I think there are also other innovations which will drive down cost:

  • There are no Radio Network Controllers (RNC) anymore, i.e. fewer network components
  • The backhaul network is radically different. While E-1/T-1 connections (cable, microwave) are still heavily used today, LTE will be rolled out with Ethernet over fiber / VDSL and microwave. Huge cost advantage here. It's not spectral efficiency operators worry about today, it's the rising E1/T1 backhaul costs.
  • In all fairness, it has to be said, that current HSPA networks are changing towards this as well in terms of backhaul and network element (e.g. one tunnel architecture) but it is not built in and the RNC is still required.
  • Another reason why LTE has a cost advantage over today's deployed networks is that technology has advanced and allows smaller base stations to be built which require less power, less space. These will be deployed from day 1 and in many cases will be put inside existing base station cabinets or mounted besides.
  •  Also count in remote radio head technology that will probably be used heavily with LTE to drive the cost down.
  • In the mid- to long term, I think LTE access will be the catalyst to have multi radio base stations with a common Ethernet based backhaul thus also driving down the cost of 2G and 3G systems to some extend that will remain in place for the time to come.

Anything else you can think of?

LTE Field Performance

Ericsson has published an interesting article in their Ericsson review (3/2008) on their latest LTE development state. Both lab and outdoor trials were done and the article together with the many graphs and pictures is an interesting read. Highly recommended! While you read, however, you should keep the following things in mind:

  • Unlike the setup recently used by T-Mobile and Nortel in Germany, only a single base station site was used, i.e. their measurement results do not reflect a typical network deployment, were neighboring cell interference will have an impact on the throughput.
  • When looking at the graphs, it should be kept in mind that according to this article by Agilent's Moray Rumney, 90% of the users will not experience a signal to noise ration (SNR) of more than 15 dB. 50% of the users will be below 5 dB. So make sure you have a look at the graphs at those locations.
  • Figure 8 shows nicely, that 64QAM modulation only makes sense at an SNR of more than 15 dB. In other words 90% of the users will not benefit from such high order modulation. However, if you can place your LTE receiver (e.g. your dongle dock) near the window in the direction of the next base station for stationary use the system will be able to server you a lot better than indoors.
  • 4×4 MIMO is nice but I doubt that we will see this implemented in base stations or real mobile devices anytime soon.

Despite these things, however, the graphs and experiences made by Ericsson should make for a nice experience in practice once LTE gets deployed and mobile devices are available.

LTE Performance Simulations

Two pointers today to performance simulations performed in 3GPP for LTE and comparison to baseline HSPA:

  • 3GPP R1-072580: Liaison statement with an overview of the results of LTE performance simulation in uplink and downlink.
  • 3GPP R1-071956: Simulations performed by Ericsson on the downlink (referenced in the document above)

The result is that 2×2 MIMO and 4×4 MIMO bring a tremendous benefit for the average cell throughput, with average cell spectral efficiency for 2×2 MIMO at 1.58 bits/s/Hz and for 4×4 MIMO at 3.04 bits/s/Hz under the same radio conditions.

Even baseline HSPA with a theoretical peak data rate of 14 MBit/s in a 5 MHz channel has a peak spectral efficiency of 2.8 bits/s/Hz which comes close to what the report say can be done only with 4×4 MIMO (who's peak spectral efficiency is even higher). So if the channel has a SNR high enough for 3 bits/s/Hz (about 8db) why doesn't basline HSPA reach this speed as well?

Hm, what am I missing? One thing might be that those users very close to the base station or an external antenna enjoying an SNR higher than 8db can push the average data rate by having a much higher transmission rate than the average. But is that alone enough for such a difference?

3G Network Sharing: How To Isolate Performance Issues Between Operators

Broadband Expert reports that 3UK and T-Mobile UK are making progress with their joint 3G radio access network and that Ericsson will manage the venture. There are different approaches of network sharing like one company covers one area while the other covers another. This approach however, sounds like one organization maintains the RAN which is then used by both mobile operators. In this scenario, I wonder how both companies will ensure that the amount of traffic and behavior of one core network does not impact the performance for customers of the other!? How will the bandwidth be shared in practice over the same base station? 3UK for example has a notoriously bad performance record and I wonder how T-Mobile will isolate itself from those issues once they start using the same base stations. I guess each company using it's own carrier(s) on a base station is probably a way to ensure this. Here's a link to 3GPP TR 22.591 which gives an overview of the available options. I wonder which ones T-Mobile and 3UK will implement with their common RAN. Also, it would be quite interesting to know if the two companies will share the RAN everywhere or just in rural areas. 5000 base stations are to be decomissioned. But where? And finally, what will happen if one company would like to densify the network in an area due to bandwidth shortage while the other can still live with the current deployment? Many questions and I am looking forward to see how this will work in practice.

Introduction to Next Generation Wireless Backhaul

As data rates and mobile Internet use is increasing one of the big challenges of mobile operators is how to keep pace with adding capacity in the backhaul network, i.e. the network that connects the base station to the rest of the network. Currently, many operators still use slow and expensive 2 MBit/s E-1 and T-1 links that don't scale well at all with the rising wireless data rates that a single base station can provide.

This article in the 3/08 edition of the Ericsson review gives a great overview of next generation IP based backhaul and how to get there. Despite only being a few pages long, it touches a lot of different topics. Here are some examples:

  • In practice, the radio access network is split in two parts: The last mile to the base station and the aggregation network to the border node to the core network. For both parts, the article explains the different technology choices.
  • For the aggregation layer, different architectures are described from VDSL, microwave Ethernet and fiber.
  • An introduction to Quality of Service, latency, jitter and circuit emulation are also not missing.
  • And finally, a migration path from current E-1 centered access networks to hybrid networks in which a base station has both E-1 and IP connectivity to full IP connectivity with pseudo wire capabilities is also inside.

Definitely, a recommended read!

3GPP Femto Specifications

The post on Femtospots a couple of days ago had some good feedback and one reader pointed me to TS 22.220 where 3GPP currently lays the ground for an end-to-end femtocell architecture, or Home NodeB architecture in 3GPP talk. Thanks for that, quite an insightful document! Here's a link to the document after the latest 3GPP meeting (December 2008) which hasn't yet made it to the official specification server. While still being a somewhat early draft today, it nevertheless gives some interesting insight into which directions operators want to go with femtos.

I've had a look at the contributors to the document and from the operator side, T-Mobile, Vodafone, Softbank, SK Telecom, and NTT-Docomo seem to be the most enthusiastic ones. On the vendor side, I've seen input from RIM, ETRI, Qualcomm, NEC, Alcatel, Huawei, Nortel and Marvel. The lists are not exhaustive but show that there is a lot of interest in the topic.

Here are the some of the highlights of the document:

Open and Closed Operation

3GPP TS 22.220 is a requirements specification so it will serve as a guideline for future stage 2 and stage 3 documents which will contain the implementation details for those requirements. So while trying to stay realistic, the document tries to explore the topic in as wide a range as possible and to keep as many options open as possible. Three operating modes are specified for femtos / Home NodeBs (UMTS) or Home eNodeBs (LTE) and I use the terms interchangeably below: The first one is called open, which means all UEs (user equipment in 3GPP talk or mobile devices) of an operator are allowed to use the cell. The second mode is called Closed Subscriber Group (CSG), which means only selected UEs, for example those belonging to a household, are allowed to use the cell. The third mode is called hybrid and combines the first two. I imagine that in hybrid mode, CSG users might potentially get higher priority and access to the local network.

Local IP Access

Speaking of local network access, the requirements specification also contains a chapter on allowing the UE access the the users home network. No specifics are mentioned yet as to how this should be implemented in practice or what kind of services could be used over such a connection. I expect that the 'how' will be clarified in stage 2 and stage 3 documents while the 'what' will be left for other standards bodies to clarify. The document says that both operator and users will have a say which users are part of the CSG and which users will be allowed to have access to local resources.

Local IP Access to the Internet

A so far empty chapter is present for how to connect to the Internet via the local network therefore bypassing the operators core network. I can hardly wait to see if this chapter will be filled with text or removed in later versions of the document.

MBMS and Mobile TV

Some parties also seem keen to use the Home NodeB for mobile TV and would like to see MBMS specified for femtos.

IMS

Further, there seem to be operators or vendors who would like to have some parts or all of an IMS implemented in the femto in a transparent way for the UE to potentially bypass the circuit switched network. I don't quite yet fully get the concept and purpose of this feature but I am sure some more text will be added to this chapter as the document evolves.

The Achilles Heel : Pre-Release 8 UEs

In my opinion the biggest overall issue for femtos used in closed subscriber group (CSG) mode is how to prevent mobiles not belonging to the CSG trying to reselect to the femto. For future 3GPP Release 8/9 compliant UEs, things can be standardized to avoid unnecessary cell reselections and signalling. TS 22.220 gives some general guidance on how that could be done by adding femto related information on the broadcast channel of the cell. For today's UEs, however, any solution has to work with what is already in place. 3GPP TR 25.280 gives a number of potential solutions in Chapter 6.2. Personally I think the Equivalent PLMN solution has a lot of merrit, but no definite recommendation of how to solve this is given yet.

German Computer Magazine measures 5.76 MBit/s in HSDPA Downlink

Edition 25/2008 of the C't, a renowned German computer magazine, contains a number of interesting articles around mobile Internet access. In one of them, 3G USB dongles have been tested and those capable of 7.2 MBit/s in downlink (HSDPA category 7/8) have reached a maximum speed of 5.76 MBit/s. Impressive, that's even higher than what I experienced myself. The test were performed on the Hanover exibition ground, where both T-Mobile and Vodafone have upgraded their 3G network and their base station backhaul to support these speeds. I assume the tests were done while no exhibition was in progress, i.e. no traffic in the cell and also no or only little traffic in other cells in the neighborhood, which means only little inter-cell interference. They also tested HSUPA and achieved uplink data rates of around 1.8 MBit/s. Again, very impressive for a live network setup.

3GPP Work Item: Multi Standard Radio (MSR)

Just a few days ago, I've speculated about GSM 'virtually' surviving for quite some time in base stations which have radio and digital modules capable of handling several air interface technologies at the same time. Looks like Ericsson has also been thinking in this direction and has recently started a Work Item in 3GPP to explore Multi Standard Radios (MSR).

The main topic of this work item is to define physical layer characteristics when one transmitter sends out several carriers. So far, all characteristics like neighbor interference and blocking where defined around a single carrier. For Multi Standard Radios, the same definitions now have to be applied around all the carriers of one base station transmitter.

Good places to look for further info are GP-081607 which contains a nice figure on how a single radio module could handle GSM, UMTS and LTE together. GP-081608 is also quite interesting, again some figures and a list of frequency bands for which MSR should be explored first. As per this paper, Ericsson would like to explore Multi Standard Radios for UMTS and LTE for the 2.1 and 2.5 GHz band (band 1 and 7), and GSM, UMTS and LTE combined radios for the 850 MHz band (US, Canada) and 900 MHz (Europe, Asia) (band 5 and 8).

The Work Item description (RP-080758) lists Alcatel-Lucent, Huawei, Nokia Siemens Networks, NTT DoCoMo, TeliaSonera and T-Mobile as supporter, so E/// is far from alone.