Many obstacles in the way of 3G coverage | Bangkok Post: tech

Providing 3G coverage will prove much more difficult than many expect and is not just a case of re-using old 1800 cells sites to the nature of the technology that could see cell sizes range from over 100 kilometres to just down the road depending on the number of active users in the cell, as Alcatel-Lucent Thailand's head of solutions and marketing, Laurent Perche, explains.

3G coverage will prove much more difficult than many expect, according to Alcatel-Lucent Thailand’s head of solutions and marketing, Laurent Perche.

Networks and coverage

In the past on GSM, it was possible to decorrelate coverage and capacity. That means putting up a big cell and when more and more users came online, the telco would put up sites in the middle of cells and carve it up. A GSM cell has a fixed size of 35 km. When the spectrum is full, the telco can simply put up new cells between existing cells, tilt the old antennas down to limit their coverage and gradually grow and evolve the network.

With 3G WCDMA (as was the case with CDMA) it is not possible to decorrelate capacity with coverage and the network has to plan both at the same time before rolling out the network. This is because CDMA (the technology) is more susceptible to interference and has a behaviour called cell breathing. A cell that has one user will be much larger than one that has 50 users, which in turn will be much larger than one with 200 users.

A user 7 km from the base station might be able to connect to a newly launched network, but just a dozen heavy data users, he would quickly fall off the edge of the cell. In town if a building was planned with five floors of office users suddenly turns into 20 floors with everyone trying to use 3G data, the cell would shrink so much that most of the people in that building would not be able to access the network.

So how many users can you squeeze into each cell? Each cell typically has three sectors (antenna banks) and a good number is 100 users per sector, so that means about 300 to 500 users per cell. Of course, if the cell is next to Pantip Plaza and everyone in there is using a USB dongle, then of course it is not going to be 100.

This means that 3G will never be able to provide the kind of coverage that GSM networks can.

Much has been said of the 100 kilometre super-cells used in Australia. This is where dual frequency networks running on 850 MHz and 2100 MHz make such a nice solution. It is possible to run an extended cell on 850 to provide coverage alongside 2100 cells that provide capacity.

Over water, without any interference, Alcatel-Lucent has managed to deploy super-cells on 2100 with a series of masts aimed at a shipping lane between Japan and Korea. The maximum range between ship and shore of that network is 150 kilometres. However, this works only because of the perfect propagation over water and the fact that there is no interference from other 3G users between the ship and the on-shore base station.

Back on land, such super-cells on 2100 would not work, even with a dedicated 5 MHz channel as interference from other 2100 networks would shrink the cell. That is why using 850/2100 (or 900/2100) is so important for rural coverage and would work only if the lower band were only used for wide-area coverage.

At the other extreme, femto cells are very low power cells that plug into a home broadband connection and route the 3G phone's data over the fixed broadband network. This would make a lot of sense for True or ToT which own both mobile and fixed line networks. It also solves the problem of in-building coverage and frees up capacity for other mobile users, giving them the marketing possibility of offering unlimited "3G" data use in their home cell.

3G technology today and tomorrow

HSPA (high speed packet access), or what people often call 3.5G, is not just about faster speeds, though that is what everyone focuses on. There are a number of other significant features.

One of the most important but underrated features is CPC (Continuous Packet Connectivity). As of the end of 2009, almost all current phones and all networks have implemented this. By keeping the connection on all the time, the main benefit is battery life. Turning on a data and turning the link up and down many times a minute, as is normal in today's always-connected phones, is not good for battery life or for network traffic.

In 3G, networks are still divided into circuit switched (voice) and packet switched (data) networks in the core but an evolution to pure packet switch networks is possible with CSoHSPA (Circuit Switched over HSPA).

CSoHSPA comes in two flavours. The first is just the radio interface part where the audio is encoded and carried by HSPA. In a way it is light version of VoIP (voice over IP) as it only takes advantage of HSPA for higher density radio transfers where the voice stream is fed into the traditional circuit switched voice once it reaches the base station.

A later version of CSoHSPA allows for full end to end VoIP and paves the way for moving to a core network that is pure IP.

"Lite" CSoHSPA is offered as part of the 3GPP specification release 7. Full end to end IP on 3G will be part of 3GPP release 8. Today, most implementations are still being rolled out on 3GPP release 6 with new tenders asking for some parts of release 7, not even the whole feature set.

Another feature of HSPA, often incorrectly referred to as HSPA+, is 64 QAM modulation. By increasing the complexity of signal processing, a single user on a 3G network that has deployed 64 QAM could get up to 21 MBPS under ideal conditions and with all the 15 codes (channels) attributed to that user.

Encoding data used to be very simple. Each frequency is a sine wave. If it was shorter, then the bit is decoded as a one; if it is longer it is a 0. With 64 QAM the number of bits encoded per sine wave theoretically goes up to 8 to 16 bits per wave.

However, while the theory is that 64 QAM will double speeds, in practice, telcos have learned that deploying 64 QAM typically gives a zero to 20 percent increase in the real world networks. However, with femto cells the increase is around 50 percent as femto cells do not have to deal with much interference or share the cell with any other user.

MIMO (multiple-in, multiple-out) antennas is another feature that has been extensively tested by Vodafone in the UK. However, in tests where cells have a mix of traditional HSPA, MIMO and 64 QAM, the real world situation is that turning on MIMO can actually decrease throughput speeds by up to 40 percent. This is because many handsets and devices are taking shortcuts in implementing 3G specifications most often for reasons of power saving. Sometimes devices already have two antennas and use one to receive and the other to transmit. When it detects that MIMO is turned on in the cell, it stops using this feature and overall speeds actually drop. MIMO also doubles the cost on the network side but does not double the capacity.

Today, MIMO and 64QAM are both available but not in the same handset at the same time as the number of antennas and complexity will mean it will be very difficult to package and thus affect the form factor. Everything is moving so fast and networks and handsets are playing catch-up.

DCHSPA (Dual Channel HSPA) will first be offered as part of 3GPP Release 8 but will use two channels on the same frequency band. In release 9, DCHSPA will offer the possibility of using two bands (900 and 2100) together to theoretically double speeds. But while DCHSPA could double speed in a laboratory environment or for a single person in a cell, if a cell has just 50 users, the benefit of DCHSPA is reduced to between nothing to 20 percent because of increased interference.

Software Defined Radio is already happening on the base station side. Alcatel-Lucent already offer a base station that can be CDMA, GSM or even LTE with just a software upgrade. However, it will have to be on the same frequency band. The next step for software defined radio is for a solution that does any standard on any band. This means that a telco could expand its CDMA on 800 today and with just a software upgrade, launch LTE 800 with the same hardware at a later date.

Long term evolution - 4G

LTE, or Long Term Evolution, often referred to as 4G is significant for many reasons. It brings economy of scale and the potential for global roaming. It also brings a full IP network and the promise of cost reduction and it brings high flexibility as potentially any IMT frequency can be used for LTE. Operators will benefit from flexible channel spacing and frequencies. All this potentially means a much reduced cost to the end user.

1G had around 50 different standards, then the world went to a 2G world with two standards, GSM and CDMA. Now it is moving to one big family that is LTE.

LTE was designed as a game changing technology. It was based on IP, packet-based with low latency and the radio part has very high modulation efficiency and throughput. However, many people thought that by that time, legacy services such as SMS would no longer be there. The challenge is how to provide legacy services to people who still want to use SMS and voice on LTE.