Simulating 6lowpan with mac802154_hwsim

Simulating 6lowpan with mac802154_hwsim

[Ramon Fontes]

Hello all,

I've been developed a wireless network emulator for IoT
(https://github.com/ramonfontes/mininet-iot) in which I'm using
mac802154_hwsim. Mininet-IoT extends Mininet-WiFi
(https://github.com/intrig-unicamp/mininet-wifi) by simulating WiFi
through mac80211_hwsim. Mininet-WiFi already supports 6lowpan with
fakelb and I moving everything to mac802154_hwsim. Since I'm using
iwpan with mac802154_hwsim and given the expertise of this community,
I was wondering whether there is any work in which simulate the data
rate based on the distance/rssi with 6lowpan. If the answer is no,
what would be the best way to do so? I was thinking of doing this with
TC, but TC wouldn't work well with multiple hops. Should I develop a
wmediumd-like simulator (https://github.com/bcopeland/wmediumd) for
mac802154_hwsim?

Best regards,
Ramon Fontes

Re: Simulating 6lowpan with mac802154_hwsim

[Alexander Aring]

On Thu, Jun 06, 2019 at 07:49:07PM -0300, Ramon Fontes wrote:
> Hello all,
> 
> I've been developed a wireless network emulator for IoT
> (https://github.com/ramonfontes/mininet-iot) in which I'm using
> mac802154_hwsim. Mininet-IoT extends Mininet-WiFi
> (https://github.com/intrig-unicamp/mininet-wifi) by simulating WiFi
> through mac80211_hwsim. Mininet-WiFi already supports 6lowpan with
> fakelb and I moving everything to mac802154_hwsim. Since I'm using
> iwpan with mac802154_hwsim and given the expertise of this community,
> I was wondering whether there is any work in which simulate the data
> rate based on the distance/rssi with 6lowpan. If the answer is no,
> what would be the best way to do so? I was thinking of doing this with
> TC, but TC wouldn't work well with multiple hops. Should I develop a
> wmediumd-like simulator (https://github.com/bcopeland/wmediumd) for
> mac802154_hwsim?
> 

rssi as 802.15.4 LQI value can be set per hop in mac802154_hwsim netlink
api. For delays/drops/etc. where is the problem using by TC netem?

LQI value is *mostly dropped to garbage right now.

It was on my list to try out netem with mac802154_hwsim but I didn't do
it yet because I have no use case.

- Alex

Re: Simulating 6lowpan with mac802154_hwsim

[Ramon Fontes]

One node (e.g. node1) can be connected simultaneously to multiple
nodes  (e.g. node2 and node3) by using only one 6lowpan interface.
That said, which node (2 or 3) will node1 taking into account? They
can have different distances and only one interface in which we can
use TC. Please correct me if I'm wrong. In my mind, the same is true
for LQI.

This code (https://github.com/ramonfontes/mininet-iot/blob/master/examples/6LoWPan.py)
produce such scenario. You can also try 6lowpan.py -m for mobility
purposes.

On Wed, Jun 12, 2019 at 11:10 AM Alexander Aring <aring@mojatatu.com> wrote:
>
> On Thu, Jun 06, 2019 at 07:49:07PM -0300, Ramon Fontes wrote:
> > Hello all,
> >
> > I've been developed a wireless network emulator for IoT
> > (https://github.com/ramonfontes/mininet-iot) in which I'm using
> > mac802154_hwsim. Mininet-IoT extends Mininet-WiFi
> > (https://github.com/intrig-unicamp/mininet-wifi) by simulating WiFi
> > through mac80211_hwsim. Mininet-WiFi already supports 6lowpan with
> > fakelb and I moving everything to mac802154_hwsim. Since I'm using
> > iwpan with mac802154_hwsim and given the expertise of this community,
> > I was wondering whether there is any work in which simulate the data
> > rate based on the distance/rssi with 6lowpan. If the answer is no,
> > what would be the best way to do so? I was thinking of doing this with
> > TC, but TC wouldn't work well with multiple hops. Should I develop a
> > wmediumd-like simulator (https://github.com/bcopeland/wmediumd) for
> > mac802154_hwsim?
> >
>
> rssi as 802.15.4 LQI value can be set per hop in mac802154_hwsim netlink
> api. For delays/drops/etc. where is the problem using by TC netem?
>
> LQI value is *mostly dropped to garbage right now.
>
> It was on my list to try out netem with mac802154_hwsim but I didn't do
> it yet because I have no use case.
>
> - Alex

Re: Simulating 6lowpan with mac802154_hwsim

[Alexander Aring]

Hi,

On Wed, Jun 12, 2019 at 02:07:26PM -0300, Ramon Fontes wrote:
> One node (e.g. node1) can be connected simultaneously to multiple
> nodes  (e.g. node2 and node3) by using only one 6lowpan interface.
> That said, which node (2 or 3) will node1 taking into account? They
> can have different distances and only one interface in which we can
> use TC. Please correct me if I'm wrong. In my mind, the same is true
> for LQI.
> 

For TC it's true, then you need to match on different addresses e.g. mac
address/short address?

For LQI this is not true, because in mac802154_hwsim you say edges as a
set of vectors (which node can reach the other node) and specify the LQI
value which the receiving node will be get.

The main difference here is that TC works on interface layer whereas
mac802154_hwsim can work on phy layer. Maybe TC can even extended for
such purpose?

> This code (https://github.com/ramonfontes/mininet-iot/blob/master/examples/6LoWPan.py)
> produce such scenario. You can also try 6lowpan.py -m for mobility
> purposes.
> 

Need to test it when I have time.

- Alex

Re: Simulating 6lowpan with mac802154_hwsim

[Alexander Aring]

Hi,

On Wed, Jun 12, 2019 at 01:20:28PM -0400, Alexander Aring wrote:
> Hi,
> 
> On Wed, Jun 12, 2019 at 02:07:26PM -0300, Ramon Fontes wrote:
> > One node (e.g. node1) can be connected simultaneously to multiple
> > nodes  (e.g. node2 and node3) by using only one 6lowpan interface.
> > That said, which node (2 or 3) will node1 taking into account? They
> > can have different distances and only one interface in which we can
> > use TC. Please correct me if I'm wrong. In my mind, the same is true
> > for LQI.
> > 
> 
> For TC it's true, then you need to match on different addresses e.g. mac
> address/short address?

After talking with the TC maintainer, he told me it's implemented as
qdisc and not as action where you probably could do something to use
netem features per neighbor as run on specifc patterns...

- Alex

Re: Simulating 6lowpan with mac802154_hwsim

[Ramon Fontes]

For LQI this is not true, because in mac802154_hwsim you say edges as a
set of vectors (which node can reach the other node) and specify the LQI
value which the receiving node will be get.

>> Ok! I just used wpan-hwsim and it seems to work fine. However, I have another question: How can I get the LQI value? I've created a monitor interface, but I couldn't observe the LQI value.

------------------
For TC it's true, then you need to match on different addresses e.g. mac
> address/short address?

After talking with the TC maintainer, he told me it's implemented as
qdisc and not as action where you probably could do something to use
netem features per neighbor as run on specifc patterns...

>> Which patterns do you mean? Can I try the mac address with TC?

Thanks a lot for your kind help.

Regards,
Ramon

On Wed, Jun 12, 2019 at 5:00 PM Alexander Aring <aring@mojatatu.com> wrote:
>
> Hi,
>
> On Wed, Jun 12, 2019 at 01:20:28PM -0400, Alexander Aring wrote:
> > Hi,
> >
> > On Wed, Jun 12, 2019 at 02:07:26PM -0300, Ramon Fontes wrote:
> > > One node (e.g. node1) can be connected simultaneously to multiple
> > > nodes  (e.g. node2 and node3) by using only one 6lowpan interface.
> > > That said, which node (2 or 3) will node1 taking into account? They
> > > can have different distances and only one interface in which we can
> > > use TC. Please correct me if I'm wrong. In my mind, the same is true
> > > for LQI.
> > >
> >
> > For TC it's true, then you need to match on different addresses e.g. mac
> > address/short address?
>
> After talking with the TC maintainer, he told me it's implemented as
> qdisc and not as action where you probably could do something to use
> netem features per neighbor as run on specifc patterns...
>
> - Alex

Re: Simulating 6lowpan with mac802154_hwsim

[Alexander Aring]

On Wed, Jun 12, 2019 at 07:52:15PM -0300, Ramon Fontes wrote:
> For LQI this is not true, because in mac802154_hwsim you say edges as a
> set of vectors (which node can reach the other node) and specify the LQI
> value which the receiving node will be get.
> 
> >> Ok! I just used wpan-hwsim and it seems to work fine. However, I have another question: How can I get the LQI value? I've created a monitor interface, but I couldn't observe the LQI value.
> 

As I told, it's mostly garbage right now. There exists some way to get
it on af802154 socket... There is currently no feature for it on
AF_PACKET or some kind of neighbor nodes database (at least upstream)
for the last one I have somehow dirty patches for it (without short
address handling which getting complex because I need kind of mapping
from other layers).

Somehow I didn't got your other message.

- Alex

Re: Simulating 6lowpan with mac802154_hwsim

[Ramon Fontes]

For TC it's true, then you need to match on different addresses e.g. mac
> address/short address?

After talking with the TC maintainer, he told me it's implemented as
qdisc and not as action where you probably could do something to use
netem features per neighbor as run on specifc patterns...

>> Which patterns do you mean? Can I try the mac address with TC?

Thanks a lot for your kind help.

Regards,
Ramon
On Wed, Jun 12, 2019 at 8:03 PM Alexander Aring <aring@mojatatu.com> wrote:
>
> On Wed, Jun 12, 2019 at 07:52:15PM -0300, Ramon Fontes wrote:
> > For LQI this is not true, because in mac802154_hwsim you say edges as a
> > set of vectors (which node can reach the other node) and specify the LQI
> > value which the receiving node will be get.
> >
> > >> Ok! I just used wpan-hwsim and it seems to work fine. However, I have another question: How can I get the LQI value? I've created a monitor interface, but I couldn't observe the LQI value.
> >
>
> As I told, it's mostly garbage right now. There exists some way to get
> it on af802154 socket... There is currently no feature for it on
> AF_PACKET or some kind of neighbor nodes database (at least upstream)
> for the last one I have somehow dirty patches for it (without short
> address handling which getting complex because I need kind of mapping
> from other layers).
>
> Somehow I didn't got your other message.
>
> - Alex

Simulating 6lowpan with mac802154_hwsim

[Ramon Fontes]

Hello all,

I've developed a wireless network emulator for IoT
(https://github.com/ramonfontes/mininet-iot) in which I'm using
mac802154_hwsim. Mininet-IoT extends Mininet-WiFi
(https://github.com/intrig-unicamp/mininet-wifi) by simulating WiFi
through mac80211_hwsim. Mininet-WiFi already supports 6lowpan with
fakelb and I moving everything to mac802154_hwsim. Since I'm using
iwpan with mac802154_hwsim and given the expertise of this community,
I was wondering whether there is any work in which simulate the data
rate based on the distance/rssi with 6lowpan. If the answer is no,
what would be the best way to do so? I was thinking of doing this with
TC, but TC wouldn't work well with multiple hops. Should I develop a
wmediumd-like simulator (https://github.com/bcopeland/wmediumd) for
mac802154_hwsim?

Best regards,
Ramon Fontes