GB2416961A

GB2416961A - Uplink status flag (USF) detection for GPRS

Info

Publication number: GB2416961A
Application number: GB0417383A
Authority: GB
Inventors: Ming Ma
Original assignee: Matsushita Electric Industrial Co Ltd
Current assignee: Panasonic Holdings Corp
Priority date: 2004-08-04
Filing date: 2004-08-04
Publication date: 2006-02-08
Anticipated expiration: 2024-08-04
Also published as: GB0417383D0; GB2416961B

Abstract

For packet based data transmissions, a base station transmits an uplink status flag (USF) to a mobile station, which USF determines whether or not the mobile station may send data in one or more uplink timeslots. In the detection methods, a mobile node is assigned an USF and detects the presence of the USF in subsequent signal bursts transmitted to the mobile node. For the GPRS coding schemes CS2-CS4, the USF detection method comprises correlating the received USF samples with a plurality of possible USFs stored at the mobile node with the correlation weighted in favour of the assigned USF. False detection of the assigned USF can be minimised by checking the unscaled correlation value against a predetermined threshold. For the GPRS coding scheme CS1, for which there is no USF pre-coding, the received USF is decoded by the mobile node to provide an indication of decision confidence and the USF detection is rejected if the decision confidence is below a threshold. The path metric value is used to minimise false detection.

Description

241 6961 Improved Uplink Status Flag Detection Method for GPRS The General

Packet Radio Service (GPRS) is a new non-voice bearer service for Global System for Mobile (GSM) communication systems that enables mobile users to have wireless access to packet data networks. In packet based data transmission, the base station first needs to communicate with the mobile station and agree on the arrangement of the timeslot(s) of the radio block to be used for data exchange. There are in total eight timeslots available per frame. The number of timeslots used depends on the multislot class of the mobile station.

Each mobile station is assigned an Uplink Status Flag (USF) and the USF flag, which is transmitted by the base station in the downlink, determines whether or not the mobile station may send data in one or more of the corresponding uplink timeslots.

GPRS employs four different coding schemes, denoted by CSI, 2, 3 and 4, in the related standard [1]. As described in [1], for the coding schemes CS2, CS3 and CS4, the first free bits (USF bits) of the data block are encoded such that the first twelve coded bits are representing the same bit pattern, irrespective of the coding scheme, depending only on the USF bits as shown in Table 1. Note that there is no USF pre-coding for CS 1. Therefore USF detection methods for CS2-4 and CS 1 are different.

Table 1 Coded USF bits for CS2-4 d(O),d(1),d(2) u(O),u(1), ,u(11) 000 000000 000 000 001 000 011 011 101 001101 110110 011 001 110 101 011 110100001 011 101 110111010110 111001 111 101 111 111 010100000 USF Detection for CS2-4: The conventional method for USF detection with CS2, 3, 4 is to calculate the correlation between received 12 bit USF code samples and stored coded USF bits shown in Table 1 and find the index corresponding to the maximum correlation. However this conventional method cannot match the performance requirement [2] with good margins. This invention provides an improved method for USF detection, which matches the performance requirement with

comfortable margins.

The invention involves a weighted correlation in favour of the USF which has already been assigned to the mobile node, as described in claim 1. An example of this method is described below: Let r,(i),i=O,I,...,II, denote the mobile node soft equaliser output (for example, in 8 bit signed format) corresponding to 12 coded USF bits, and u(i,m),i=O,l,...,ll,m=0,1,

.,7, denote the coded USF pattern shown in Table 1. The USF detection algorithm first calculates the following eight correlation values: J(m)= , ,,tr(i)x(2u(i,m)- 1), 0 < m < 7 (1) =o where the parameter axon is defined as follows: l3, Al > 1, if m = assigned _ USF Ofn, = 1.0, else (2) Therefore, in the decision, the assigned USF value is favoured in order to increase correct USF detection rate. In general, by increase the "ill" value for the assigned USF value we increase the performance margin in USF detection test. However at the same time this will also decrease the performance margin in USF false alarm test when the input is random RF or a random USF value that is not equal to the assigned USF value, which is defined by the section 6.4 item g of the standard GSM 05.05 [2]. This value needs to be optimised. In our implementation, I]= 1.25 will give both good USF correct detection rate and false alarm rate...DTD: Based on equation ( l) and (2), the detected_USF is obtained according to the following: detected _USF = index{m, maximize J(m),O < m < 7} (3) if detected_USF is equal to assigned_USF, we report that the assigned USF is detected.

In order to prevent false decision, check the unscaled correlation value corresponding the assigned_USF_value: J(assigned _ USF),, = J(assigned _ USF)/,l3 (4) Where ''u'' denotes the unscaled value.

In general, if for random RF input or a random USF is not equal to the assigned_USF value, this unscaled correlation value J(assigned_USF)u will become much less than the normal case, so if J(assigned _ USF),, < correlation _ threshold ( we do not report the assigned_USF is detected even the detected_USF is equal to assigned_USF in order to prevent false alarm. In our implementation, correlation_threshold = gives a good false alarm performance.

To further improve false alarm, the averaged unscaled correlation value corresponding to assigned_USF over several USF detection period (for example, over two detection period, i.e. average over current and previous detection) can be used: I N-l-l J(assigned _ USF), = N J(assigned _USF)k (6) where J(assigned_USF)k is the unscaled correlation value corresponding to the assigned_USF for kth USF detection. This averaged value can then be compared with correlation_threshold in order to decide whether or not report the assigned_USF is detected.

In our implementation, using averaged correlation over two continuous USF detections gives a good false alarm performance, which matches the standard requirement with good margin.

USF Detection for CS1: For CSI, there is no USF pre-coding, and the first three bits in the convolutionally decoded data determine the detected USF value. If the detected_USF is equal to the assigned_USF, then we report that the assigned USF is detected. In the conventional method for detecting CSI there is no provision for false alarm performance. We therefore propose to address this point by using decision confidence data frm the mobile node decoder, as described in claim 6.

An example of this method is described below: However for random RF input or random USF which is not equal to the assigned_USF, the path metric reported by convolutional decoder will be used to prevent false alarm.

In general, if for random RF input or a random USF is not equal to the assigned_USF value, the path metric reported from our convolutional decoder will become much larger than normal case, so if path _metric metric _threshold (7) we do not report the assigned_USF is detected even if the detected_USF is equal to assigned_USF in order to prevent false alarm. In our implementation, metric_threshold = 32767 gives a good false alarm performance, however different implementation, in particular convolutional decoder, will have different metric threshold To further improve false alarm, the averaged path metric over several USF detection period can be used: I N-/-l path _metric =- path _metrick where path_metrick is the path metric reported from convolutional decoder for kth USF detection, then compare this averaged value with path_threshold and decide whether or not report the assigned_USF is detected. In our implementation, using averaged path metric over four continuous USF detections gives a good false alarm performance, which matches the standard requirement with good margin.

Reference: [1] 3GPPTS05.03, Channel Coding, V8.7.0.

[2] 3GPP TS05.05, Radio Transmission and Reception, V8.16.0 s

Claims

Claims 1. A method of detecting an uplink flag (USF) in a communications

system in which: a plurality of possible USFs are stored at a mobile node, one of the flags is allocated to the mobile node in a message transmitted to the mobile node, and the mobile node detects the presence of the USF in subsequent signal bursts transmitted to the mobile node, the detection method comprising correlating received USF samples with the stored USFs with the correlation being weighted in favour of the assigned USF.
2. A method as claimed in claim 1 in which the correlation is weighed in favour of the assigned USF by a factor of 1.25.
3. A method as claimed in claim 1 or 2 in which the USFis determined as the one of the stored possible USF values for which the correlation is greatest, subject to a correlation factor exceeding a predetermined threshold.
4. A method as claimed in claim 1, 2 or 3 in which the correlation is repeated over a plurality of samples to provide an averaged correlation value.
5. Use of a method as claimed in any preceding claim in a GPRS communications system to detect coded uplink status flags decoded according to any of the standard coding schemes CS2, CS3 or CS4.
6. A method of detecting an uplink status flag (USF) in a communications system in which: a USF is allocated to the mobile node in a message transmitted to the mobile node, and the mobile node detects the presence of the USF in subsequent signal bursts transmitted to the mobile node, wherein the USF is decoded by the mobile node in a decoder which provides an indication of decision confidence and the USF detection is rejected if the decision confidence is below a predetermined threshold.
7. A method as claimed in claim 6 in which the USF is part of a transmitted convolutional code decoded by a convolutional decoder and the decision confidence indication is the path metric.
8. A method as claimed in any preceding claim in which the path metric averaged over a plurality of samples is used in the threshold comparison.
9. A method as claimed in any preceding claim in which at least one additional USF flag is allocated to the mobile node and the same detection method is used for each flag.