Messages in this thread | | | Date | Sat, 29 Jan 2000 12:09:39 -0600 | From | watermodem <> | Subject | Re: Auto-Adaptive scheduler - Final chapter ( the numbers ) ... |
| |
BTW... As a brain teaser next time you place a call with a cell phone consider the foot print of your simple action. Also, when you see these numbers consider the optimization required to achieve acceptable service. IN ADDITION consider that most of the times mentioned here are actually spent waiting on a VLR or HLR associated with a swich to look up the id for that cellphone and setup the billing record. (Accross the board looking at all switch providers 600 to 800 ms seems typical for this simple act.)
In terms of hardware touched WORST CASE: (average is 10 or 20 cpus)
PDC (japan tdma): you could touch upto 2450 processors just to setup your call. This includes dsps, land line, etc... And it takes under a second.
CDMA (depends on company and quality of service) This one is a wild estimate as one can handover while setting up and be in N-Way where N is determined by your ablity to pay... but worst case about 5000 processors is a good estimate. If the provider has single rate vocoders 1.2 to 1.8 seconds is a usual call setup time. If the provider is mixed mode (ie mixes say lucent and motorola and samsung equipment and has various vocoding rates) 1.6 to 2.5 is a usual call setup time.
GSM (simple form without data) Worst case about 500 processors. .6 to 1.2 seconds
This ignores the O&M aspect that can and does bring many more processors to bear. O&M tends to be customer specific with many prefering to roll their own so estimates here are iffy.
Data protocols such as IP only add to the complexity. Of course if IP wasn't billed it would be the simpler one.
I am making no point here just providing some food for thought.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |