This post is by Avijit Ghosh, assistant vice president of technology, Aricent
The ‘Self-organizing Networks’ (SON) concept is a ‘hot’ topic for today’s wireless networks, especially with the expected proliferation of small cells and heterogeneous networks.
The idea is that the network should, while minimising cost and staying within constraints that may be applicable, automatically and continually adjust itself to maximise its own key performance indicators: generally coverage, capacity & quality of experience.
It is also expected that such a system should generate information, analysis and visualisation to assist effective forward planning of the service provider’s business.
SON & Network Management
A wireless network is meant to provide connectivity to mobile endpoints over a designated geographical area using one or more designated slices of the radio spectrum. To accomplish this, the network deploys radio transceiver equipment at ‘cell sites’ spread over this geographical area.
Once the network is deployed and starts providing service, it will be possible to perform actual measurements of its performance. These measurements can then be used to adjust the system. This makes it a ‘closed loop’ process.
Changing the behavior of any installed piece of equipment (within its capability range) can be accomplished by adjusting its configuration parameters.
A self-organising network is one where these tasks are automated, in some form of feedback-control system.
Considering only the radio access part of these networks, the picture above seeks to show how this can happen. The lines in red are the ‘traditional’ paths of information flow, while the lines in blue are the ‘new’ way of doing things with SON. As SON becomes more and more sophisticated, the red lines should become less and less ‘important’—culminating at a point where they are exercised only in exceptional circumstances.
While SON algorithms implement some form of feedback-control logic, they are distinct from other feedback-control loops that also exist in a radio network—such as power and timing control loops that operate at millisecond to second timescales, or radio resource management algorithms that operate at a granularity of several seconds to minutes. As with all network management, SON algorithms will take action infrequently (tens of minutes to hours…to even days apart), and will (with the exception of reacting to failure) observe (i.e. collect data) for long durations before acting.
As it is understood today, SON encompasses the following types of tasks:
- Plug & Play installation
- Automatic neighbor relations
- Mobility robustness optimization
- Coverage and capacity optimization
- Contention access (e.g. RACH) optimization
- Energy saving
- Mobility load balancing
The SON concept applies equally well to all wireless technologies—those operating in licensed bands, such as LTE, UMTS/HSPA & GSM/EDGE (and even CDMA/EVDO & WiMAX), and also those operating in unlicensed bands, such as Wi-Fi.
However, the benefits become more compelling for small cells, which are expected to be deployed in large numbers. Driven by the demand for mobile broadband, LTE, UMTS/HSPA and Wi-Fi technologies are more likely to see small-cell and heterogeneous network deployments.
When it comes to involving the terminal in the SON process (to augment drive tests and to pre-empt customer complaints) however, standardisation of the air interface is essential. This is happening for the LTE and UMTS/HSPA air interfaces—but not for the GSM/EDGE air-interface. This also isn’t in place for the Wi-Fi air interface.