Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project!

## RE: [Playerstage-developers] Gazebo Truth

 RE: [Playerstage-developers] Gazebo Truth From: Jason Redi - 2003-11-06 16:02:02 ```Hi Nathan (and all), > Also, we should find out what other wifi simulators are out there and > see if we can incorporate any of those efforts into Gazebo. The Big Three simulators that people tend to use are Opnet, NS-2, and Qualnet. Opnet and Qualnet are major commercial efforts (\$\$\$ for = companies, but usually free for non-profits). Industry and military organizations tend to like Opnet and to a smaller extent, Qualnet, instead of NS-2 = because of their more detailed approaches to modeling RF and radio reception, as well as the momentum of doing models in those simulators for many years. Academia usually prefers NS-2, because it's open source and many other academic efforts have been done using NS-2 for years so there are models = you can get to compare your protocol against. My personal feeling is that trying to tie a network/RF simulator into a robotics/physics simulator may be overkill for many applications/experiments. An alternative for just determining whether = two nodes can communicate would be something like: - Calculate line of sight (LOS) path loss using 2-ray ground=20 reflection model for particular frequency (2.4Ghz), antenna=20 gain (probably 0dBi) and antenna height (short or tall robot). - Determine polygons of foliage, buildings, etc. =20 - Determine if LOS line intersects any polygons and add additional attenuation for loss (e.g. Weissburger model for foliage, etc). - Result is total path loss between two nodes. (PL) - One you have the pathloss, you can use the transmit power (30mW?),=20 receive antenna gain (probably 0dBi) and receiver sensitivity=20 to determine if the packet got the receiver uncorrupted. Note that receiver sensitivity is usually provided as the number where = the receiver will see *something*, but not necessarily be able to decode the packet out of noise. So you would probably add another 20dB to receiver sensitivity to say if a packet was received uncorrupted. (As an = example, the intersil chipset has something like -92dBm receiver sensitivity, but = we have done some experiments that say you need -72dBm to receive 1500byte packets, uncorrupted, 99.999% of the time. If someone wants to map the = SNR to modulation curve and feed that into a CRC check, you can probably = figure that relation out directly). If you want to simulate multipath, you can do it by expanding the = two-ray ground reflection model (two-ray =3D=3D ground ray + LOS ray) to be a = six-ray model (ground + LOS + ceiling + right + left + behind). Each of the = rays can constructively or destructively interfere with each other because = the different lengths create different phase shifts at the receiver. = Depending on how polygons are represented in your simulator, this might not be so = hard to determine these rays. The next step after that would be to calculate the noise (and possible collisions from other nodes transmitting in the area). Then, if you = want to simulate the MAC layer interactions, which further limit throughput, you could add the whole RTS/CTS/data/ACK mechanism in there. These last two steps add significant complexity over just determining whether you can communicate at all between two nodes. Whether this is worth it, depends quite a lot on your application. What do you think? Jason --- Dr. Jason K. Redi E: redi@... BBN Technologies, a part of Verizon V: 617-873-4236 10 Moulton St., Cambridge, MA, 02138 F: 617-873-2092=20 > -----Original Message----- > From: Nathan M. Farrington [mailto:nfarring@...]=20 > Sent: Wednesday, November 05, 2003 6:28 PM > To: Pranav > Cc: playerstage-developers@...;=20 > bpg@...; redi@... > Subject: Re: [Playerstage-developers] Gazebo Truth >=20 >=20 > Hi Pranav, >=20 > I remember Ben Grocholsky was interested in > simulating wifi in Gazebo. Have you talked to him? >=20 > Our group is also interested in a wifi simulation. The=20 > question is now, > to what extent do you want to simulate it? For instance, we are using > BBN's ad-hoc network routing software developed by Jason Redi > . Ideally, we would like to simulate not only=20 > the 802.11b > radios we are using, but also BBN's ad-hoc networking algorithms. >=20 > If you just want to simulate a regular 802.11b (or g) radio, then the > task is much simpler, but still tricky depending on what exactly you > want. For simulating the signal strength of the radio, you=20 > could use an > empirical formula based on a model of 2.4GHz communications. For > instance, the signal strength has a sharp dip at a distance where > multipath reflections off of the ground destructively=20 > interfere with the > line-of-sight transmission from the transmitter to the receiver. The > problem should be more pronounced when there are more objects=20 > to reflect > off of, especially inside of a tunnel. >=20 > An improvement would be to drop packets in proportion to the=20 > degradation > in signal strength, to the point where robots can no longer=20 > communicate > with each other! And what about collisions caused by multiple=20 > radios in > the same area? I'm not an expert in networking or wireless networking > like Dr. Redi, but I imagine that wireless network simulation has > numerous issues. Anyway, its not an easy problem but I think=20 > it would be > very useful to lots and lots of people. >=20 > Also, we should find out what other wifi simulators are out there and > see if we can incorporate any of those efforts into Gazebo. >=20 > Regards, >=20 > Nathan Farrington >=20 >=20 > ----- Original Message ----- > From: Pranav > Date: Wednesday, November 5, 2003 11:42 am > Subject: [Playerstage-developers] Gazebo Truth >=20 > > Hi guys, > >=20 > > Just exploring the possibility of having some kind of wifi=20 > > simulation ( a > > bare-bones one ) in gazebo. For that i think i would need to know=20 > > all the > > other wifi devices ( which i would have to write) in the world..=20 > > anyonehave any ideas on this? > >=20 > > thanks > > p > >=20 > >=20 > > ------------------------------------------------------- > > This SF.net email is sponsored by: SF.net Giveback Program. > > Does SourceForge.net help you be more productive? Does it > > help you create better code? SHARE THE LOVE, and help us help > > YOU! Click Here: http://sourceforge.net/donate/ > > _______________________________________________ > > Playerstage-developers mailing list > > Playerstage-developers@... > > https://lists.sourceforge.net/lists/listinfo/playerstage-developers > >=20 >=20 >=20 >=20 ```

### Thread view

 Re: [Playerstage-developers] Gazebo Truth From: Nathan M. Farrington - 2003-11-05 23:28:50 ```Hi Pranav, I remember Ben Grocholsky was interested in simulating wifi in Gazebo. Have you talked to him? Our group is also interested in a wifi simulation. The question is now, to what extent do you want to simulate it? For instance, we are using BBN's ad-hoc network routing software developed by Jason Redi . Ideally, we would like to simulate not only the 802.11b radios we are using, but also BBN's ad-hoc networking algorithms. If you just want to simulate a regular 802.11b (or g) radio, then the task is much simpler, but still tricky depending on what exactly you want. For simulating the signal strength of the radio, you could use an empirical formula based on a model of 2.4GHz communications. For instance, the signal strength has a sharp dip at a distance where multipath reflections off of the ground destructively interfere with the line-of-sight transmission from the transmitter to the receiver. The problem should be more pronounced when there are more objects to reflect off of, especially inside of a tunnel. An improvement would be to drop packets in proportion to the degradation in signal strength, to the point where robots can no longer communicate with each other! And what about collisions caused by multiple radios in the same area? I'm not an expert in networking or wireless networking like Dr. Redi, but I imagine that wireless network simulation has numerous issues. Anyway, its not an easy problem but I think it would be very useful to lots and lots of people. Also, we should find out what other wifi simulators are out there and see if we can incorporate any of those efforts into Gazebo. Regards, Nathan Farrington ----- Original Message ----- From: Pranav Date: Wednesday, November 5, 2003 11:42 am Subject: [Playerstage-developers] Gazebo Truth > Hi guys, > > Just exploring the possibility of having some kind of wifi > simulation ( a > bare-bones one ) in gazebo. For that i think i would need to know > all the > other wifi devices ( which i would have to write) in the world.. > anyonehave any ideas on this? > > thanks > p > > > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program. > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/ > _______________________________________________ > Playerstage-developers mailing list > Playerstage-developers@... > https://lists.sourceforge.net/lists/listinfo/playerstage-developers > ```
 Re: [Playerstage-developers] Gazebo Truth From: john sweeney - 2003-11-06 00:15:42 ```For research I have been doing, I have modified versions of stage and the ns2 network simulator to run in (some sort of) synchronization. This allows me to use ns to simulate wireless traffic, using its built-in wireless models, but the locations of the wireless nodes come from the locations of the robots, which is controlled by stage. It's a pretty big hack, and it's on older versions of stage and ns (1.3.2 and 2.1b9a respectively), and probably wicked overkill for most robotics work, but if there's interest I could put it somewhere for people to look at. john Nathan M. Farrington wrote: > Hi Pranav, > > I remember Ben Grocholsky was interested in > simulating wifi in Gazebo. Have you talked to him? > > Our group is also interested in a wifi simulation. The question is now, > to what extent do you want to simulate it? For instance, we are using > BBN's ad-hoc network routing software developed by Jason Redi > . Ideally, we would like to simulate not only the 802.11b > radios we are using, but also BBN's ad-hoc networking algorithms. > > If you just want to simulate a regular 802.11b (or g) radio, then the > task is much simpler, but still tricky depending on what exactly you > want. For simulating the signal strength of the radio, you could use an > empirical formula based on a model of 2.4GHz communications. For > instance, the signal strength has a sharp dip at a distance where > multipath reflections off of the ground destructively interfere with the > line-of-sight transmission from the transmitter to the receiver. The > problem should be more pronounced when there are more objects to reflect > off of, especially inside of a tunnel. > > An improvement would be to drop packets in proportion to the degradation > in signal strength, to the point where robots can no longer communicate > with each other! And what about collisions caused by multiple radios in > the same area? I'm not an expert in networking or wireless networking > like Dr. Redi, but I imagine that wireless network simulation has > numerous issues. Anyway, its not an easy problem but I think it would be > very useful to lots and lots of people. > > Also, we should find out what other wifi simulators are out there and > see if we can incorporate any of those efforts into Gazebo. > > Regards, ```
 Re: [Playerstage-developers] Gazebo Truth From: Richard Vaughan - 2003-11-06 01:05:39 ```I'm very interested in what you guys achieve in this direction. For example, if you come up with a reasonably clean Stage/ns2 interface, let's talk about how to make this available to everyone. Wei Ye and I did this with Stage (back when it was called `Arena') and it worked pretty well. We achieved de-facto syncronisation by slowing down Arena's clock a bit so ns2 ran just about fast enough. Not pretty, but enough to show that the parameters of the wireless propagation model had significant effects on our robots' overall behavior. Wireless is a significant part of a robot's environment these days, so having decent models would be very useful. Paper in ICRA'01. Richard. On Wed, 5 Nov 2003, john sweeney wrote: > For research I have been doing, I have modified versions of stage and > the ns2 network simulator to run in (some sort of) synchronization. > This allows me to use ns to simulate wireless traffic, using its > built-in wireless models, but the locations of the wireless nodes come > from the locations of the robots, which is controlled by stage. > > It's a pretty big hack, and it's on older versions of stage and ns > (1.3.2 and 2.1b9a respectively), and probably wicked overkill for most > robotics work, but if there's interest I could put it somewhere for > people to look at. > > john > > Nathan M. Farrington wrote: > > Hi Pranav, > > > > I remember Ben Grocholsky was interested in > > simulating wifi in Gazebo. Have you talked to him? > > > > Our group is also interested in a wifi simulation. The question is now, > > to what extent do you want to simulate it? For instance, we are using > > BBN's ad-hoc network routing software developed by Jason Redi > > . Ideally, we would like to simulate not only the 802.11b > > radios we are using, but also BBN's ad-hoc networking algorithms. > > > > If you just want to simulate a regular 802.11b (or g) radio, then the > > task is much simpler, but still tricky depending on what exactly you > > want. For simulating the signal strength of the radio, you could use an > > empirical formula based on a model of 2.4GHz communications. For > > instance, the signal strength has a sharp dip at a distance where > > multipath reflections off of the ground destructively interfere with the > > line-of-sight transmission from the transmitter to the receiver. The > > problem should be more pronounced when there are more objects to reflect > > off of, especially inside of a tunnel. > > > > An improvement would be to drop packets in proportion to the degradation > > in signal strength, to the point where robots can no longer communicate > > with each other! And what about collisions caused by multiple radios in > > the same area? I'm not an expert in networking or wireless networking > > like Dr. Redi, but I imagine that wireless network simulation has > > numerous issues. Anyway, its not an easy problem but I think it would be > > very useful to lots and lots of people. > > > > Also, we should find out what other wifi simulators are out there and > > see if we can incorporate any of those efforts into Gazebo. > > > > Regards, > > > > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program. > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/ > _______________________________________________ > Playerstage-developers mailing list > Playerstage-developers@... > https://lists.sourceforge.net/lists/listinfo/playerstage-developers > > -- Richard Vaughan / HRL Laboratories / vaughan@... / +1 310.317.5689 ```
 Re: [Playerstage-developers] Gazebo Truth From: john sweeney - 2003-11-06 02:05:42 ```In my case I extended stage's shared memory to include its clock time (or maybe that was there already? I forget exactly) and then I let ns share it as well. They keep in sync by waiting for each other to catch up to whatever simulated time they have. I also use this shared memory for swapping parameters between the simulators. As far as the interface, it's not the best that could be done given that model, I think, but with more thought it might be made pretty clean... john Richard Vaughan wrote: > I'm very interested in what you guys achieve in this direction. For > example, if you come up with a reasonably clean Stage/ns2 interface, let's > talk about how to make this available to everyone. Wei Ye and I did this > with Stage (back when it was called `Arena') and it worked pretty well. We > achieved de-facto syncronisation by slowing down Arena's clock a bit so > ns2 ran just about fast enough. Not pretty, but enough to show that the > parameters of the wireless propagation model had significant effects on > our robots' overall behavior. Wireless is a significant part of a robot's > environment these days, so having decent models would be very useful. > Paper in ICRA'01. > ```
 RE: [Playerstage-developers] Gazebo Truth From: Josh Bers - 2003-11-06 14:55:45 ```Hi, I work with Jason Redi on enhancing ad-hoc networks for use with = robots. We have developed an extended interface for sharing status of the ad-hoc network (topolgy, link costs, etc) with a robot controller. So far we = have built an SNMP MIB and Java client access to the info. We are thinking = about building a Player interface/driver to the data as well. Would there be interest in this? Josh > -----Original Message----- > From: playerstage-developers-admin@...=20 > [mailto:playerstage-developers-admin@...]=20 > On Behalf Of Nathan M. Farrington > Sent: Wednesday, November 05, 2003 6:28 PM > To: Pranav > Cc: playerstage-developers@...;=20 > bpg@...; redi@... > Subject: Re: [Playerstage-developers] Gazebo Truth >=20 >=20 > Hi Pranav, >=20 > I remember Ben Grocholsky was=20 > interested in simulating wifi in Gazebo. Have you talked to him? >=20 > Our group is also interested in a wifi simulation. The=20 > question is now, to what extent do you want to simulate it?=20 > For instance, we are using BBN's ad-hoc network routing=20 > software developed by Jason Redi . Ideally, we=20 > would like to simulate not only the 802.11b radios we are=20 > using, but also BBN's ad-hoc networking algorithms. >=20 > If you just want to simulate a regular 802.11b (or g) radio,=20 > then the task is much simpler, but still tricky depending on=20 > what exactly you want. For simulating the signal strength of=20 > the radio, you could use an empirical formula based on a=20 > model of 2.4GHz communications. For instance, the signal=20 > strength has a sharp dip at a distance where multipath=20 > reflections off of the ground destructively interfere with=20 > the line-of-sight transmission from the transmitter to the=20 > receiver. The problem should be more pronounced when there=20 > are more objects to reflect off of, especially inside of a tunnel. >=20 > An improvement would be to drop packets in proportion to the=20 > degradation in signal strength, to the point where robots can=20 > no longer communicate with each other! And what about=20 > collisions caused by multiple radios in the same area? I'm=20 > not an expert in networking or wireless networking like Dr.=20 > Redi, but I imagine that wireless network simulation has=20 > numerous issues. Anyway, its not an easy problem but I think=20 > it would be very useful to lots and lots of people. >=20 > Also, we should find out what other wifi simulators are out=20 > there and see if we can incorporate any of those efforts into Gazebo. >=20 > Regards, >=20 > Nathan Farrington >=20 >=20 > ----- Original Message ----- > From: Pranav > Date: Wednesday, November 5, 2003 11:42 am > Subject: [Playerstage-developers] Gazebo Truth >=20 > > Hi guys, > >=20 > > Just exploring the possibility of having some kind of wifi > > simulation ( a > > bare-bones one ) in gazebo. For that i think i would need to know=20 > > all the > > other wifi devices ( which i would have to write) in the world..=20 > > anyonehave any ideas on this? > >=20 > > thanks > > p > >=20 > >=20 > > ------------------------------------------------------- > > This SF.net email is sponsored by: SF.net Giveback Program. Does=20 > > SourceForge.net help you be more productive? Does it > > help you create better code? SHARE THE LOVE, and help us help > > YOU! Click Here: http://sourceforge.net/donate/=20 > > _______________________________________________ > > Playerstage-developers mailing list=20 > > Playerstage-developers@... > > https://lists.sourceforge.net/lists/listinfo/playerstage-developers > >=20 >=20 >=20 >=20 >=20 >=20 > ------------------------------------------------------- > This SF.net email is sponsored by: SF.net Giveback Program.=20 > Does SourceForge.net help you be more productive? Does it > help you create better code? SHARE THE LOVE, and help us help > YOU! Click Here: http://sourceforge.net/donate/=20 > _______________________________________________ > Playerstage-developers mailing list=20 > Playerstage-developers@... > https://lists.sourceforge.net/lists/listinfo/playerstage-developers >=20 ```
 RE: [Playerstage-developers] Gazebo Truth From: Jason Redi - 2003-11-06 16:02:02 ```Hi Nathan (and all), > Also, we should find out what other wifi simulators are out there and > see if we can incorporate any of those efforts into Gazebo. The Big Three simulators that people tend to use are Opnet, NS-2, and Qualnet. Opnet and Qualnet are major commercial efforts (\$\$\$ for = companies, but usually free for non-profits). Industry and military organizations tend to like Opnet and to a smaller extent, Qualnet, instead of NS-2 = because of their more detailed approaches to modeling RF and radio reception, as well as the momentum of doing models in those simulators for many years. Academia usually prefers NS-2, because it's open source and many other academic efforts have been done using NS-2 for years so there are models = you can get to compare your protocol against. My personal feeling is that trying to tie a network/RF simulator into a robotics/physics simulator may be overkill for many applications/experiments. An alternative for just determining whether = two nodes can communicate would be something like: - Calculate line of sight (LOS) path loss using 2-ray ground=20 reflection model for particular frequency (2.4Ghz), antenna=20 gain (probably 0dBi) and antenna height (short or tall robot). - Determine polygons of foliage, buildings, etc. =20 - Determine if LOS line intersects any polygons and add additional attenuation for loss (e.g. Weissburger model for foliage, etc). - Result is total path loss between two nodes. (PL) - One you have the pathloss, you can use the transmit power (30mW?),=20 receive antenna gain (probably 0dBi) and receiver sensitivity=20 to determine if the packet got the receiver uncorrupted. Note that receiver sensitivity is usually provided as the number where = the receiver will see *something*, but not necessarily be able to decode the packet out of noise. So you would probably add another 20dB to receiver sensitivity to say if a packet was received uncorrupted. (As an = example, the intersil chipset has something like -92dBm receiver sensitivity, but = we have done some experiments that say you need -72dBm to receive 1500byte packets, uncorrupted, 99.999% of the time. If someone wants to map the = SNR to modulation curve and feed that into a CRC check, you can probably = figure that relation out directly). If you want to simulate multipath, you can do it by expanding the = two-ray ground reflection model (two-ray =3D=3D ground ray + LOS ray) to be a = six-ray model (ground + LOS + ceiling + right + left + behind). Each of the = rays can constructively or destructively interfere with each other because = the different lengths create different phase shifts at the receiver. = Depending on how polygons are represented in your simulator, this might not be so = hard to determine these rays. The next step after that would be to calculate the noise (and possible collisions from other nodes transmitting in the area). Then, if you = want to simulate the MAC layer interactions, which further limit throughput, you could add the whole RTS/CTS/data/ACK mechanism in there. These last two steps add significant complexity over just determining whether you can communicate at all between two nodes. Whether this is worth it, depends quite a lot on your application. What do you think? Jason --- Dr. Jason K. Redi E: redi@... BBN Technologies, a part of Verizon V: 617-873-4236 10 Moulton St., Cambridge, MA, 02138 F: 617-873-2092=20 > -----Original Message----- > From: Nathan M. Farrington [mailto:nfarring@...]=20 > Sent: Wednesday, November 05, 2003 6:28 PM > To: Pranav > Cc: playerstage-developers@...;=20 > bpg@...; redi@... > Subject: Re: [Playerstage-developers] Gazebo Truth >=20 >=20 > Hi Pranav, >=20 > I remember Ben Grocholsky was interested in > simulating wifi in Gazebo. Have you talked to him? >=20 > Our group is also interested in a wifi simulation. The=20 > question is now, > to what extent do you want to simulate it? For instance, we are using > BBN's ad-hoc network routing software developed by Jason Redi > . Ideally, we would like to simulate not only=20 > the 802.11b > radios we are using, but also BBN's ad-hoc networking algorithms. >=20 > If you just want to simulate a regular 802.11b (or g) radio, then the > task is much simpler, but still tricky depending on what exactly you > want. For simulating the signal strength of the radio, you=20 > could use an > empirical formula based on a model of 2.4GHz communications. For > instance, the signal strength has a sharp dip at a distance where > multipath reflections off of the ground destructively=20 > interfere with the > line-of-sight transmission from the transmitter to the receiver. The > problem should be more pronounced when there are more objects=20 > to reflect > off of, especially inside of a tunnel. >=20 > An improvement would be to drop packets in proportion to the=20 > degradation > in signal strength, to the point where robots can no longer=20 > communicate > with each other! And what about collisions caused by multiple=20 > radios in > the same area? I'm not an expert in networking or wireless networking > like Dr. Redi, but I imagine that wireless network simulation has > numerous issues. Anyway, its not an easy problem but I think=20 > it would be > very useful to lots and lots of people. >=20 > Also, we should find out what other wifi simulators are out there and > see if we can incorporate any of those efforts into Gazebo. >=20 > Regards, >=20 > Nathan Farrington >=20 >=20 > ----- Original Message ----- > From: Pranav > Date: Wednesday, November 5, 2003 11:42 am > Subject: [Playerstage-developers] Gazebo Truth >=20 > > Hi guys, > >=20 > > Just exploring the possibility of having some kind of wifi=20 > > simulation ( a > > bare-bones one ) in gazebo. For that i think i would need to know=20 > > all the > > other wifi devices ( which i would have to write) in the world..=20 > > anyonehave any ideas on this? > >=20 > > thanks > > p > >=20 > >=20 > > ------------------------------------------------------- > > This SF.net email is sponsored by: SF.net Giveback Program. > > Does SourceForge.net help you be more productive? Does it > > help you create better code? SHARE THE LOVE, and help us help > > YOU! Click Here: http://sourceforge.net/donate/ > > _______________________________________________ > > Playerstage-developers mailing list > > Playerstage-developers@... > > https://lists.sourceforge.net/lists/listinfo/playerstage-developers > >=20 >=20 >=20 >=20 ```