OpenBSCOpenBSCHandover » History » Version 2
ipse, 02/19/2016 10:47 PM
Link to the handover improvement proposal.
1 | 1 | laforge | = Handover code in OpenBSC = |
---|---|---|---|
2 | |||
3 | Both the osmo-nitb and osmo-bsc programs implement hand-over between cells in the same BSC (intra-BSC). |
||
4 | |||
5 | We implement the algorithms recommended in the book ''Performance Enhancements in a Frequency Hopping GSM Network'' |
||
6 | (by Thomas Toftegaard Nielsen and Jeroen Wigard). |
||
7 | |||
8 | The actual implementation is in http://cgit.osmocom.org/cgit/openbsc/tree/openbsc/src/libbsc/handover_decision.c |
||
9 | |||
10 | There are some tunable parameters but in the end it's not all that |
||
11 | complicated. The BSC tells the phone the list of neighbor BTS ARFCN, |
||
12 | the phone reports in every SACCH uplink frame the Rx Level measurements |
||
13 | of those neighbors. |
||
14 | |||
15 | There are a number of different reasons: |
||
16 | 1. interference HO, if the Rx level > -85dBm but the !RxQual is bad over more then 3 out of 4 measurements |
||
17 | 1. bad quality HO, if the !RxQual is bad in 3 out of 4 measurements |
||
18 | 1. low level HO, if the Rx level is < - 110dBm |
||
19 | 1. distance HO, if the TA is above a threshold |
||
20 | 1. power budget HO, i.e. if we have better cells |
||
21 | |||
22 | the decision is based on an average !RxLev over a sliding window of |
||
23 | measurements, taking into account a configurable hysteresis level. |
||
24 | |||
25 | We have tested it in a couple of field tests (HAR2009, 27C3 and so on), |
||
26 | but none of our big real-world deployments is currently using hand-over, |
||
27 | as they typically have small single-BTS sites that have no neighboring |
||
28 | cells. |
||
29 | 2 | ipse | |
30 | = Handover improvements = |
||
31 | |||
32 | A proposal for the handover improvement is published [wiki:handover here]. |