210 likes | 319 Views
Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [ Adaptive Frequency Hopping - An instant channel replacement approach for both ACL and SCO ] Date Submitted: [September, 2001]
E N D
Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [Adaptive Frequency Hopping - An instant channel replacement approach for both ACL and SCO] Date Submitted: [September, 2001] Source: [H. Gan, V. Sapozhnykov, B. Treister, E. Skafidas, et. al.] Company [Bandspeed Inc.] Address [Level 9, 500 Collins Street, Melbourne, Victoria, Australia] Voice:[61 3 9614 6299 , FAX: [61 3 9614 6699] E-Mail:[h.gan, b.treister, v.sapozhnykov@bandspeed.com.au] Re: [A new simple approach for adaptive frequency hopping] Abstract: [This document describes a new simple approach for adaptive frequency hopping, an instant channel replacement to intelligently use bad channels in the hopping sequence] Purpose: [Introducing a new approach for adaptive frequency hoping to include in 802.15.2] Notice: This document has been prepared to assist the IEEE P802.15. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein. Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE and may be made publicly available by P802.15. H. Gan, V. Sapozhnykov, et. al.
Adaptive Frequency Hopping (AFH) - A Simple Instant Channel Replacement Approach for both ACL and SCO Hongbing Gan, Vitaliy Sapozhnykov, Bijan Treister, Stan Skafidas, et. al. Bandspeed Inc. H. Gan, V. Sapozhnykov, et. al.
Definition of Channel Pair A channel pair is comprised of two channels: • First channel, Master Tx/Slave Rx channel, at even-numbered timeslot • Second channel, Slave Tx/Master Rx, ie, Slave return channel, at odd-numbered timeslot Channel Pair Channel Pair Master Tx Rx Tx Rx f1 f3 f2 f4 Slave Rx Tx Rx Tx Odd-numbered Timeslot Even-numbered Timeslot Odd-numbered Timeslot Even-numbered Timeslot H. Gan, V. Sapozhnykov, et. al.
Definitions • N: Total number of hopping channels • Nmin: Minimum number of channels to be used, set by regulations such as FCC • G: Good channel • B: Bad channel • BN: Bad channel to be removed legally from the hopping sequence • BK: Bad channel to keep in the hopping sequence • NG, NB, NBN, NBK: Number of good channel, Number of bad channel, Number of bad channel to remove, Number of bad channel to keep, respectively • N = NG + NB • NB = BK + BN H. Gan, V. Sapozhnykov, et. al.
AFH Covers All Possible Channel Classification Scenarios • Scenario 1: • BN= 0, Bk > 0, all Bad channels are kept, the AFH intelligently use BK in the new hopping sequence • Scenario 2: • BN> 0, BK > 0, the AFH replaces BN with good channels G or BK, and intelligently use BK in the new hoping sequence • Scenario 3: (Ideal scenario) • BN> 0, BK= 0, all Bad channels are replaced with Good channels in the new hopping sequence H. Gan, V. Sapozhnykov, et. al.
AFH Covers All the Traffic Types H. Gan, V. Sapozhnykov, et. al.
AFH Approach 1 - Standard Instant Channel Replacement (Standard ICR) (Details in IEEE 802.15 Document 01/435r0, 01/438r0) H. Gan, V. Sapozhnykov, et. al.
G B G B G G B B G B B G B G B G Principle of Standard ICR Pass Re-transmission Blocked Blocked Original New Pass Pass Blocked Blocked • Original ‘Good Good’ and ‘Bad Bad’ channel pairs are kept in their original positions in the hopping sequence • ‘Good Bad’ channel pairs are instantly replaced to ‘Good Good’ channel pairs • ‘Bad Good’ channel pairs are instantly replaced to ‘Bad Bad’ channel pair • Throughput improved due to newly created ‘Good Good’ channel pairs H. Gan, V. Sapozhnykov, et. al.
Principle of Standard ICR For each channel pair, starting from Master Tx channel: • Removing BN by replacing it with a randomly selected G or BK, to maintain equal usage of G and BK • Whenever Master Tx channel is OR replaced to G, replace the Slave return channel to G if it is BN or BKoriginally, to form a ‘G G’ channel pair and secure a transaction • Whenever Master Tx channel is OR replaced to BK, replace the Slave return channel to BKif it is BN or Goriginally, to form a ‘BK BK’ channel pair, to remove BN and save a usage of G • Channel replacement on a per channel pair basis H. Gan, V. Sapozhnykov, et. al.
Standard ICR Processing Table In case Master TX BN replaced with a G In case Master TX BN replaced with a BK • When NO BK, Case 1, 3 ,4, 6 are processed • When NO BN, Case 1, 2, 10, 11 are processed H. Gan, V. Sapozhnykov, et. al.
G G G G G G G BN BN G G G BK G BK BN G BK G BK G BN G G G BN BK BK BK G G G G G BK G G G G BK BK G BN BN BN BK BK G G BK BK G G BK G G Example Portion of Original and AFH Hopping Sequence AFH In this example, 7 more ‘Good Good’ channel pairs are created H. Gan, V. Sapozhnykov, et. al.
Flowchart A: Used for Master Tx/Slave Rx timeslots, i.e., Even-numbered timeslot H. Gan, V. Sapozhnykov, et. al.
Flowchart B: Used for Slave Tx/Master Rx timeslots, i.e., Odd-numbered timeslot H. Gan, V. Sapozhnykov, et. al.
AFH Approach 2 - ‘Fit Best’ Instant Channel Replacement (Fit Best ICR) H. Gan, V. Sapozhnykov, et. al.
General Principle of Fit Best ICR • For HV2 and HV3: • Fit the best channels to channel pairs of the reserved SCO Timeslots • For mixed SCO + ACL: • Fit the best channels to channel pairs of the reserved SCO Timeslots, if SCO has higher priority • Fit the best channels to channel pairs of the ACL Timeslots, if ACL has higher priority • Use Standard ICR if neither ACL nor SCO has higher priority H. Gan, V. Sapozhnykov, et. al.
T T T T T T T T T T T T R R R R R R R R R R R R Mixed SCO (HV2, HV3) and ACL Links HV2 ACL HV2 ACL HV2 ACL HV2 reserves half Timeslots, transmits every four Timeslots HV3 ACL ACL HV3 ACL ACL HV3 reserves one third Timeslots, transmits every six Timeslots H. Gan, V. Sapozhnykov, et. al.
How Fit Best ICR works • For channel pairs at higher priority timeslots: • Case 1: ‘Good Good’ ‘Good Good’ • Case 2: ‘Good Bad’ ‘Good Good’ • Case 3: ‘Bad Good’‘Good Good’ • Case 4: ‘Bad Bad’ ’Good Good’ • For every Case 3 replacement, a Good Channel Usage Debt (GUD) Counter is incremented by 1; For every Case 4 replacement, the GUD Counter is incremented by 2. • In lower priority timeslots, if there is GUD debt, a ‘Good Bad’ channel pair will be replaced to ‘Bad Bad’ channel pair to repay the GUD by 1; A ‘Good Good’ channel pair will be replaced to ‘Bad Bad’ channel pair to repay the GUD by 2. • GUD balance is always towards ZERO, to maintain equal channel usage. H. Gan, V. Sapozhnykov, et. al.
How Fit Best ICR works • Example: HV2 + ACL, HV2 Voice Link has higher priority (shown in next slide) • In HV2 slots, do the following replacement: • ‘Good Good’ channel pairs are kept as usual • ‘Good Bad’ replaced to ‘Good Good’ as usual • ‘Bad Good’ replaced to ‘Good Good’, increment GUD debt counter by 1 • ‘Bad Bad’ replaced to ‘Good Good’, increment GUD debt counter by 2 • In ACL slots, do the following replacements • If any GUD Debt, ‘Good Good’ channel pairs are replace to ‘Bad Bad’, decrease GUD debt counter by 2 • If any GUD Debt, ‘Good Bad’ replaced to ‘Bad Bad’, decrease GUD debt counter by 1 • ‘Bad Good’ replaced to ‘Bad Bad’ as usual • ‘Bad Bad’ are kept as usual • If ACL has higher priority, just do the reverse. H. Gan, V. Sapozhnykov, et. al.
G G BK G BK BK G G G G BK G G BK G BK BK G G BK G BK BK BK G G G G G BK BK BK G G G BK BK G G G BK BK G G BK G BK G G BK G G G BK G G How Fit Best ICR works(Assuming HV2 has higher priority) ACL HV2 ACL HV2 ACL HV2 HV2 1 2 3 4 5 6 7 ACL HV2 ACL HV2 ACL HV2 ACL 8 9 10 11 12 13 14 AFH ACL HV2 ACL HV2 ACL HV2 HV2 1 2 3 4 5 6 7 ACL HV2 ACL HV2 ACL HV2 ACL 8 9 10 11 12 13 14 H. Gan, V. Sapozhnykov, et. al.
Fit Best ICR Processing Table At timeslots for higher priority traffic, such as voice HV2 At timeslots for lower priority traffic H. Gan, V. Sapozhnykov, et. al.
Traffic Type VS. AFH Method H. Gan, V. Sapozhnykov, et. al.