I came across another new acronym last week (mobile telecoms is full of them of course), where LTE standards have “borrowed” a term from the computing industry. Self-CHOP isn’t a new karate move or a new type of TV dinner, but relates to the significantly complex automatic management of LTE cellsites. I’d argue that many of these features are already available in commercial 3G femtocells, so their macrocell cousins will be playing catchup for LTE (the next generation 4G mobile phone standard).
Self-CHOP is borrowed from the computing world, where autonomic computers are:
- Self Configuring
- Self Healing
- Self Optimising
- Self Provisioned
Let’s look at each of these in turn
Currently, network planning engineers (with the aid of sophisticated predictive radio planning tools) produce huge datasets with millions of parameters that are downloaded into the basestations. Some relate to physical attributes, such as antenna direction/tilt, frequencies, power levels etc. while others relate to internal software parameters such as cell identity, location area and handover threshold levels.
Today’s femtocells automatically determine many of these values. There are network wide parameters, such as which frequencies are available to use but parameters are typically not finely tuned on a per femtocell basis by central planners. The list of nearby cells which calls can be handed over to – called the neighbour list - is automatically determined by network listen to see what is nearby.
When a cellsite goes offline, ideally its neighbours would crank up the power levels to expand their coverage area to fill the gap. A feature in femtocells today (called Network Listen), scans the environment for the presence of other 2G and 3G cells. Some vendors ha ve designed their femtocells to scan the environment frequently and can adapt their power levels to accommodate changes in the external environment, effectively increasing their range when appropriate. This could also include updating the neighbour list if a nearby cell disappeared/went offline.
A further expansion of this theme is to communicate with nearby femtocells to measure and automatically adapt to changing conditions. For example, if everybody in a building congregates in one area then the potentially overloaded cell can reduce its range and transfer excess traffic to nearby femtocells.
Femtocells may also be aware of potential issues which might cause problems such as a continuous ping-pong handover between the femtocell and outside macrocell. In this case optimisation might take the form of choosing the algorithm or thresholds being used.
The “plug and play” nature of residential femtocells is a far cry from the typical macrocell configuration requiring manual design and configuration by network engineers. Such self provisioning ensures that all femtocells have the latest operating software,are automatically allocated unique and appropriate cell identifiers, area codes etc.
LTE researchers are working on these ideas
I came across this Ericsson presentation about self-CHOP (undated) which suggested some of the ideas being considered for LTE. This is an academic university thesis so should not imply any specific approaches being implemented by Ericsson or any other LTE vendor in this respect.
The suggestion is that all LTE cellsites are GPS equipped and report their location to a central database. This allows a list of nearby cellsites within a set range to be identified, from which neighbour lists can be generated. It also allows the centralised and automated allocation of LTE cell identifiers (there are only ~500 unique codes available) so that they don’t overlap.
A recent article in the IET magazine (Institution of Engineering and Technology - a UK professional body) discussed the topic further. Some of the current radio optimisation tool vendors suggested that there is still scope for centralised aspects of the self-CHOP - unlike many of today's 3G femtocells it can't all be done locally by each unit.
Perhaps some 3G femtocell designers can benefit from these concepts?
I suspect that it’s the macrocell designers who could learn most from current commercially available self-CHOP algorithms and techniques used in femtocells than the other way around. But it would be an oversight for those developing 3G femtocells not to stay abreast of the concepts being investigated and adopted for LTE. I expect this will be rich ground for patent filings all round.
And there’s much closer interworking between LTE macro and femtocells
Today’s 3G femtocells are pretty much independent of the outdoor macrocells, designed to be more or less invisible by operating at very low power, indoors and/or at different frequencies. With a few exceptions for very important customers, unusual situations and large enterprise/buildings, the macrocell outdoor network won’t be reconfigured for individual femtocell installations.
With LTE however, the macro network also has this inherent self-CHOP capability.
This means that LTE macrocells and femtocells will interwork together much more closely than for 3G to enable a full self-optimising network. As you plug in your LTE femtocell, the outdoor LTE macrocell may be reconfigured to adapt and work harmoniously. Radio planners may be uncomfortable with the thought that millions of LTE femtocells around the country can directly communicate and adapt the configuration of the macrocell network in almost real-time. This may still be at the theoretical stage (you can see that the standards for LTE Self Optimising Networks is still pretty thin), but is the intended path of the future.
3G4G blog: Long list of reading material on LTE
Finland Technical University: That Ericsson presentation from an the academic thesis
Nomor: LTE newsletter on Self Organising Networks (from 2008)