Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Scheduling of downlink HSDPA data onto the Iub

IP.com Disclosure Number: IPCOM000031560D
Original Publication Date: 2004-Sep-29
Included in the Prior Art Database: 2004-Sep-29
Document File: 5 page(s) / 28K

Publishing Venue

Motorola

Related People

Steve Barrett: AUTHOR

Abstract

A method is discussed by which HSDPA MAC_d PDU's are scheduled onto the backhaul by the RNC in such a way that system throughput is maximised whilst avoiding the complexities associated with flow control based solutions.

This text was extracted from a Microsoft Word document.
This is the abbreviated version, containing approximately 43% of the total text.

Scheduling of downlink HSDPA data onto the Iub

by Steve Barrett

Abstract

A method is discussed by which HSDPA MAC_d PDU's are scheduled onto the backhaul by the RNC in such a way that system throughput is maximised whilst avoiding the complexities associated with flow control based solutions.

Introduction

The paper addresses a method that can be implemented in a 3GPP UMTS Terrestrial Radio Access Network (UTRAN), Release 5 and above, that supports High Speed Downlink Packet Access (HSDPA).  The technology may also be applicable to other radio networks where there is a Base Site Controller (BSC) that is connected through a limited bandwidth channel to a Base Transceiver Site (BTS), and where the BTS includes an air interface scheduling function.

Problem(s) To Be Solved

It is necessary to ensure that the Node B HSDPA scheduler has as large a number of UE's to select between as possible so that it can maximise system throughput through a process known as fade riding.  Fade riding involves determining which UE's are currently under going the best channel conditions and then scheduling those UE's in preference to others.  Fade riding can result in many dB's of capacity improvement.   However, in order to maximise the gains of fade riding it is necessary that wherever possible the Node B should have buffered packets on as many of the bearers/flows as possible - so as to increase the scheduler's chance of scheduling traffic to those UE's undergoing the very best conditions.  

The 3GPP standards define signalling for a flow control mechanism.  In this mechanism the Node B is informed of whether there are packets buffered at the RNC for a given flow and is also informed of what the priority is of those packets (this is done using Capacity Request messages).  The Node B can then send Capacity Allocation messages which grant the RNC a number of capacity credits / ie a quantity of packets that the RNC is allowed to forward to the Node B for the given flow.

Advantages of Node B flow control:

1) The Node B knows what its buffer fill status is, and from which MAC_d flows it would benefit in receiving more packets.  It also knows what the current radio conditions are on each link, this information could also potentially be used in determining which PDU’s should be forwarded since if a link is undergoing good conditions the scheduler may prefer to schedule more packets on that link.  

Disadvantages of Node B flow control

1)  In order to benefit from using the information as listed above (such as buffer fill / link quality info),  the Node B should grant relatively small parcels of capacity credit to the RNC.  Small parcels of credit should be allocated since eg radio conditions / buffer fill status will change and if a big grant has been made then the Node B won’t be able to exploit the new conditions.   This in turn will require frequent messaging, sending of Capacity Allocation and Capacity Request messages consuming backhaul capacity a...