When generating soft decisions from symbols, what's the best decision function to use? Is some linear approximation of a sigmoid suitable? What effect does the decision function have on the performance of the system?
soft decisions: what decision function to use?
Started by ●July 6, 2010
Reply by ●July 6, 20102010-07-06
On Jul 6, 2:16�pm, "gct" <smcallis@n_o_s_p_a_m.gmail.com> wrote:> When generating soft decisions from symbols, what's the best decision > function to use? �Is some linear approximation of a sigmoid suitable? �What > effect does the decision function have on the performance of the system?I think you need to be more specific in your question. I'm not sure if "sigmoid" here refers to the lower part of one's colon or to a general class of mathematical functions or to a specific sigmoid function such as the logistic.
Reply by ●July 6, 20102010-07-06
>On Jul 6, 2:16=A0pm, "gct" <smcallis@n_o_s_p_a_m.gmail.com> wrote: >> When generating soft decisions from symbols, what's the best decision >> function to use? =A0Is some linear approximation of a sigmoid suitable?=>=A0What >> effect does the decision function have on the performance of thesystem?> >I think you need to be more specific in your question. I'm not sure if >"sigmoid" here refers to the lower part of one's colon or to a general >class of mathematical functions or to a specific sigmoid function such >as the logistic. >I was referring to something of the general form 1/(1+e^-t)
Reply by ●July 6, 20102010-07-06
gct wrote:> When generating soft decisions from symbols, what's the best decision > function to use?By definition, soft decision = likelihood of this symbol. The particular likelihood function depends on the signal and the channel.> Is some linear approximation of a sigmoid suitable?It depends.> What effect does the decision function have on the performance of the system?Yes, it does. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Reply by ●July 6, 20102010-07-06
On 07/06/2010 11:25 AM, Clay wrote:> On Jul 6, 2:16 pm, "gct"<smcallis@n_o_s_p_a_m.gmail.com> wrote: >> When generating soft decisions from symbols, what's the best decision >> function to use? Is some linear approximation of a sigmoid suitable? What >> effect does the decision function have on the performance of the system? > > I think you need to be more specific in your question. I'm not sure if > "sigmoid" here refers to the lower part of one's colon or to a general > class of mathematical functions or to a specific sigmoid function such > as the logistic.But that does put a whole new spin on "I pulled it out of my ass". -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by ●July 6, 20102010-07-06
>By definition, soft decision = likelihood of this symbol. The particular >likelihood function depends on the signal and the channel. > >Vladimir Vassilevsky >DSP and Mixed Signal Design Consultant >http://www.abvolt.com >How about a simple BPSK signal over an AGWN channel?
Reply by ●July 6, 20102010-07-06
On 07/06/2010 01:07 PM, gct wrote:>> By definition, soft decision = likelihood of this symbol. The particular >> likelihood function depends on the signal and the channel. >> >> Vladimir Vassilevsky >> DSP and Mixed Signal Design Consultant >> http://www.abvolt.com >> > > How about a simple BPSK signal over an AGWN channel?For any signal over any channel, just calculate the a-posteriori likelihood that the transmitted bit was a 1. That's your soft decision. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" was written for you. See details at http://www.wescottdesign.com/actfes/actfes.html
Reply by ●July 6, 20102010-07-06
On Jul 6, 9:19�pm, Tim Wescott <t...@seemywebsite.com> wrote:> On 07/06/2010 01:07 PM, gct wrote: > > >> By definition, soft decision = likelihood of this symbol. The particular > >> likelihood function depends on the signal and the channel. > > >> Vladimir Vassilevsky > >> DSP and Mixed Signal Design Consultant > >>http://www.abvolt.com > > > How about a simple BPSK signal over an AGWN channel? > > For any signal over any channel, just calculate the a-posteriori > likelihood that the transmitted bit was a 1. �That's your soft decision.Have a look at Sayed's adaptive filters, chapter one. -Momo
Reply by ●July 6, 20102010-07-06
On Jul 6, 3:14�pm, Tim Wescott <t...@seemywebsite.com> wrote:> On 07/06/2010 11:25 AM, Clay wrote: > > > On Jul 6, 2:16 pm, "gct"<smcallis@n_o_s_p_a_m.gmail.com> �wrote: > >> When generating soft decisions from symbols, what's the best decision > >> function to use? �Is some linear approximation of a sigmoid suitable? �What > >> effect does the decision function have on the performance of the system? > > > I think you need to be more specific in your question. I'm not sure if > > "sigmoid" here refers to the lower part of one's colon or to a general > > class of mathematical functions or to a specific sigmoid function such > > as the logistic. > > But that does put a whole new spin on "I pulled it out of my ass".Absolutely! Furthermore, because of the uncertainty principle, the spin must be quantized. Greg
Reply by ●July 7, 20102010-07-07
