RADIO NETWORK CONTROLLER DIMENSIONING The need for high speed, high bandwidth data drives the evolution of wireless networks from 2G to 3G. The radio network controller is one of the key network elements in the 3G wireless network as it handles both circuit switched and packet switched data and also the associated signaling in both domains. Therefore the dimensioning of this network element, in terms of control processors, network processors and addressing the terminating link capacities is crucial from the operator’s point of view. This white paper describes an approach for the dimensioning of the radio network controller in the 3G wireless network. It also explains the RNC architecture and configurations through examples. WHITE PAPER SRIVATSA K N
16
Embed
RADIO NETWORK CONTROLLER · PDF fileRADIO NETWORK CONTROLLER DIMENSIONING The need for high speed, high bandwidth data drives the evolution of wireless networks from
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
RADIO NETWORKCONTROLLER
DIMENSIONINGThe need for high speed, high bandwidth data drives theevolution of wireless networks from 2G to 3G. The radio networkcontroller is one of the key network elements in the 3G wirelessnetwork as it handles both circuit switched and packet switcheddata and also the associated signaling in both domains. Thereforethe dimensioning of this network element, in terms of controlprocessors, network processors and addressing the terminatinglink capacities is crucial from the operator’s point of view.
This white paper describes an approach for the dimensioning ofthe radio network controller in the 3G wireless network. It alsoexplains the RNC architecture and configurations throughexamples.
INTRODUCTIONThere are various models which have the ability to predict the traffic requirements in a 3G network. But given a traffic model,the next step is to dimension the network according to the parameters of the traffic model. In this paper, an approach ispresented for the dimensioning of the RNC.
The dimensioning of RNC can be viewed from 3 different aspects:• User plane requirements – what is the required bandwidth for supporting both CS and PS traffic• Control plane capacities – what are the required control processor capacities for handling both CS and PS signaling
across all interfaces• Link capacities – what are the link capacity requirements across all interfaces
Given these aspects, this paper discusses approaches dealing with:• Computation of user plane capacities required across all the interfaces in terms of network processor requirements• Computation of control plane capacities across all the interfaces in terms of control processor requirements• Computation of link capacities required across all the interfaces
This information will be useful for designing the RNC in terms of determining the number of control/network processorsrequired and the corresponding link capacities required at each interface based on the estimated subscriber traffic in that area.Another usage could be that of validating an existing design of the RNC with the dimensions given by the approach. Thisinformation can also be used for existing RNCs to check if the current dimensions of the RNC are inline with the traffic offeredon a per RNC basis. While scaling up to meet increased bandwidth requirements, the new required RNC configuration can bedetermined.
ARCHITECTUREIn UMTS, RNC is part of the UTRAN (Universal Terrestrial Radio Access Network). A UTRAN consists of a set of radio networksubsystems, which in turn consist of one RNC and multiple Node-Bs. The Node-B connects to the RNC though the Iub interface.One RNC connects to another RNC through the Iur interface, and the RNC is connected to the MSC through the Iu-CS interfaceand to the SGSN through the Iu-PS interface.
Figure 1: High level network architecture of UMTS networks
All elements in a 3G network connect through ATM or IP networks, where the throughput, QoS, and link configurations can bemanaged at a granular level. At the physical layer, the elements can be connected through E1/T1 links or SONET/STM links orEthernet links. The Iub, Iu-CS, and Iu-PS interfaces can be either ATM based, IPoATM based or fully IP based.
THE DIMENSIONING APPROACHThis dimensioning approach will be discussed at the following levels:
• Inputs from the traffic model
• The BHCA computations that are done at the user and signalling planes based on the inputs from the traffic model
• Computation of the actual user plane requirements
• Computation of the actual control plane requirements
• Computation of link capacities
A visual representation of this approach is given in Figure 2.
The parameters that constitute the traffic model can be broadly classified under call attempts, duration of calls and the datarates associated with the calls. Under these categories, the parameters need to be input are given in the following tables (theseparameters are for a single user).
*Values in italics are the ones directly used in the user/control plane computations
Table 4: BHCA computations for control plane
User plane
For the user plane, only the peak user traffic that will need to be processed is required. The main BHCA parameters that need tobe computed based on the traffic model are given below:
• Total user traffic per session on the uplink/downlink• Total user traffic per person per busy hour• Total traffic (UL/DL) for the entire RNC per BH
User Plane Requirements
The calculation of the user plane bandwidth requirements directly comes from the input Traffic Model. In general, the usertraffic associated with umts can be classified into conversational, streaming, interactive and background.
For each of these traffic classes, there can be a separate traffic model. Ultimately when the user plane capacity needs to becomputed an aggregate of all these separate traffic models needs to be considered.
The steps involved in calculating the user plane requirements are briefly summarized hereafter:
Step 1:Step 1:Step 1:Step 1:Step 1: The basic traffic model inputs are entered. For example, parameters are number of subscribers, % of subscribersengaged in CS, no. of CS/PS mobile originated, call attempts per user/hour etc.
Step 2: Step 2: Step 2: Step 2: Step 2: Based on the input Traffic model parameters in step 1, some parameters related to the user plane requirements arecomputed. For example computed parameters are no. of successful CS/PS mobile originated call attempts per user/hr, no. ofsuccessful CS MTCA per user/hr etc.
Step 3: Step 3: Step 3: Step 3: Step 3: Finally, the required bandwidth on the UL and DL for user plane traffic is computed. This is done for each assumed trafficclass (e.g., CS, PS-64K, PS-128K and PS-256K).
The approach to computing the control plane requirements is as follows:
• Using the traffic model, determine the number of primary events (see below) occurring during a BH (busy hour)• For each such event, determine the number of messages to be handled at each of the RNC interfaces (Iub, Iur, Uu, Iu)• Estimate the complexity of handling each message using a method which determines the number of fields that need to
be processed in each message and the complexity of these fields. This is done using the Message Analysis Method• The complexity derived from each message is converted to LOC based on a multiplication factor (which is computed
based on some existing implementation of telecom scenarios related to 2G networks)• The total number of LOC executed per scenario is determined by multiplying the LOC obtained in the previous step with
the estimated number of occurrences of that scenario (this is derived from the traffic model)• Finally, the estimated LOC/sec is aggregated on a protocol/interface basis to give the estimated processing requirements
for each control processor. In this case, aggregation is made on NBAP(Iub), RRC(Uu), RNSAP(Iur) and RANAP(Iu)protocol basis.
Message complexity analysis method – a summaryMessage complexity analysis method – a summaryMessage complexity analysis method – a summaryMessage complexity analysis method – a summaryMessage complexity analysis method – a summary
This method primarily involves analysing the complexity of incoming/outgoing messages at each interface based on the fieldspresent in the message and the nature of these fields (basic, compound etc.).
A sample output of this method is given below, where the complexity of the messages URA Update and URA Update Confirmare determined at the Uu interface. Also the number of LOC executed per second as a result of processing these messages is alsogiven.
RRC Messages UCE Number ofMessages
InterfaceUCE
ACE LOC Numberof Occurrences
(per sec)
Total LOCexecuted(per sec)
URA Update 8 1
URA UpdateConfirm
20 1 28 30.24 3156 5334.3361.764
• Unadjusted complexity estimate (UCE) – This is obtained from the message complexity analysis• Interface UCE – This is just a summation of the complexity estimates of the messages passing through that interface• Adjusted complexity estimate (ACE) – This is the result of adjusting the user interface UCE with the adjustments factor
derived from the general system characteristics determined for RNC• LOC – This is the result of converting the ACE to LOC using the multiplicative factor• Number of occurrences – This is obtained from the Traffic Model• Total LOC executed – LOC multiplied by Number of occurrences per second gives the Total LOC executed for that
Control plane requirements – a sample outputControl plane requirements – a sample outputControl plane requirements – a sample outputControl plane requirements – a sample outputControl plane requirements – a sample output
Based on the above approaches, the following was computed for the signalling plane requirements:
Figure 3: Control plane requirements – sample output
Link Capacities Computation
As a final step, the link capacities at each interface and the corresponding network processor requirements will be computed.This is based on the following inputs:
• Total user traffic as determined in Error! Reference source not found.Error! Reference source not found.Error! Reference source not found.Error! Reference source not found.Error! Reference source not found.• Total signaling traffic will be taken as a percentage of this entire user traffic (for the purposes of this paper, it is taken as
percentage of the traffic, depending on the interface).• The total of the user and signaling traffic would yield the actual traffic estimated at the RNC level• For each interface, the traffic can be split as follows:
- IuB – The entire traffic generated from CS and PS calls and the associated signaling would be taken into accounthere.
- IuCS - The traffic is further split into CS and PS traffic (based on the traffic model). The traffic associated with CS willbe used for determining the capacity at this interface.
- IuPS – Similar to IuCS, the traffic associated with PS will be used.- Iur – Here the entire data traffic (CS+PS) is taken into account and a percentage of this is estimated to flow through
this interface.• Based on the above bandwidth requirements, the link capacities (in terms of number of E1 or OC-3 links) can be derived
Here, the assumption is that the load is evenly spread across all the processors – with an assumed upper threshold of 75%occupancy. Beyond this threshold there may be performance issues in dealing with real time traffic. Therefore, the aboveconfiguration diagram is only intended to provide the number of CPUs required and not give the occupancy of each CPU.
Since the processors used for RRC and RNSAP are relatively underutilized, they can also be used for supporting OAM relatedmodules. Typically these functionalities would include:
• Support of interface towards the OMC-R
• Support of interface towards local terminals
• Supervision and maintenance of all modules within the RNC
• Performance measurement functions
• Maintenance of central disk-backed database (or any other equivalent data storage base)
* Network Processor – Intel IXP, Control Processor – Intel Pentium III Xeon
Table 6: Sample RNC specifications
CONCLUSIONIn this white paper, an approach to dimensioning the RNC was discussed. The main aspects dealt with were the computation ofuser plane requirements, control plane requirements and the link capacity requirements. This approach would prove to beuseful from the RNC designer’s perspective when the expected traffic model is known. Also from the operator’s perspective thisinformation can be used while considering the impacts related to scaling up the network to meet increased user demands.
It has to be kept in mind that this is an iterative approach. Initially, based on the number of subscribers that needs to besupported, the model generates a sample RNC configuration. These outputs have to be validated against hardware constraints,budget constraints and the estimated cost-benefit ratio.
Based on the results of this validation, the desired RNC specifications can be used as inputs to the model to derive the subscriberbase that can be supported. These steps can be performed iteratively to generate the most optimum specifications in terms ofcost, benefits and hardware constraints.
CONCLUSIONIn this white paper, an approach to dimensioning the RNC was discussed. The main aspects dealt with were the computation ofuser plane requirements, control plane requirements and the link capacity requirements. This approach would prove to beuseful from the RNC designer’s perspective when the expected traffic model is known. Also from the operator’s perspective thisinformation can be used while considering the impacts related to scaling up the network to meet increased user demands.
It has to be kept in mind that this is an iterative approach. Initially, based on the number of subscribers that needs to besupported, the model generates a sample RNC configuration. These outputs have to be validated against hardware constraints,budget constraints and the estimated cost-benefit ratio.
Based on the results of this validation, the desired RNC specifications can be used as inputs to the model to derive the subscriberbase that can be supported. These steps can be performed iteratively to generate the most optimum specifications in terms ofcost, benefits and hardware constraints.
ACRONYMS
BHCA Busy Hour Call Attempts
Iub Interface between Node-B and RNC
Iur Interface between RNC and RNC
Iu-CS Interface between RNC and Circuit Switched Network
ABOUT THE AUTHORSrivatsa is a Project Leader in Wipro Technologies with about 6 years experience in the telecom domain and 4 years in thewireless domain. He has been involved in the specification, design, implementation and testing of various features related toGSM/GPRS base station controller. He holds a bachelor’s degree in Computer Science. His areas of interest include aspectsinvolving migration of existing wireless 2G systems to 3G systems and integration of wireless systems (GPRS/WLAN).
ABOUT WIPRO TECHNOLOGIESWipro is a PCMM Level 5 and SEI CMMi Level 5 certified global IT Services company providing comprehensive IT solutions andservices (including systems integration, IS outsourcing, package implementation, software application development andmaintenance) and Research and Development services (hardware and software design, development and implementation) tocorporations all over the world. Wipro’s unique value proposition is further delivered through their pioneering offshoreoutsourcing Model and stringent Quality Processes of SEI and Six Sigma.
WIPRO IN TELECOMWipro Telecom Solutions Division offers comprehensive solutions for telecommunication to confront challenges, and convertevery challenge into an opportunity. With over two decades of telecom experience, Wipro Telecom solutions offers a widerange of solutions in wireless networking, broadband (data, optical and access networking), voice switching, networkmanagement and hardware design.
Wipro also provides several IPs, components and reference solutions in these domains which help our clients in saving costsand providing the time-to-market advantage. We offer complete consulting, architecting, design, implementation andmaintenance services to the telecom equipment manufacturers.