Is aodv a wireless sensor network protocol?

[silsila...@gmail.com]

I have 2 doubts.

1) Aodv is an adhoc on demand protocol. My doubt is whether aodv can be employed as a wireless sensor network routing protocol??

2) Can the concept of base station be used in wireless adhoc networks as a station (or desination) receiving messages from the peers and having more processing power?

Whether the concept of base station in wireless sensor networks can be used here in case of adhoc networks too??

Please clarify these

Thanks

[Tommaso Pecorella]

Hi, the reply is in-line.

On Saturday, July 19, 2014 6:17:16 PM UTC+2, silsila...@gmail.com wrote:

I have 2 doubts.

1) Aodv is an adhoc on demand protocol. My doubt is whether aodv can be employed as a wireless sensor network routing protocol??

Sure thing. If you want to drain your WSN battery in no time. The man issue may be another one tho. AODV ns-3 implementation is IPv4 only, and if you try to say that your WSN is running IPv4... well, be ready to be a walking joke. "Hey, look, it's the one who wanted to use IPv4 on a WSN. Hahahahaha."

Seriously, AODV is not suitable for a WSN. AODV is made for networks with medium / high mobility, while a WSN are mostly static.

 

2) Can the concept of base station be used in wireless adhoc networks as a station (or desination) receiving messages from the peers and having more processing power?

 I don't see why not.

Whether the concept of base station in wireless sensor networks can be used here in case of adhoc networks too??

The problem may be another one: there's no "Base Station" in a WSN either. There's a "Gateway", or a "Router", but there's not a "Base Station".

In a WSN the MAC / PHY protocols are typically multihop and ad-hoc. As a consequence there's no "Base Station". There's usually a node with more power handling some network tasks, like routing, bridging to WiFi or Ethernet, MAC address distribution, etc., but it's not a Base Station.

Cheers,

T.

[silsila...@gmail.com]

Thank you !!!


Following is a routing framework proposed for my thesis work. Could you please check it and identify the problems associated when developing such a protocol??

I have taken Aodv as the base protocol.Since Aodv is not trusted i want to include trust. For that, in addition to routing able , another "Neighborhood table (NT)" is proposed wherein the neighbours of a particular node as well as trust value associaed with them will be stored.

In order to get information about trust of each node, Base station broadcasts are used. Base station will be broadcasting some packet (which contain information about undelivered sequence numbers) in certain intervals. These packets will be analyzed by the node and will calculate trust of neighbors using a methodology.(I am not explaining it here).

I also added a recommendation mechanism where if a node does not have trust information about a neighbor,it may contact  3 other neighbors asking for the trust .For that a new request should be multi casted. And  3 replies from these neighbors will be used by the former node to calculate the unknown trust.

Further if a node calculates trust of a neighbor based on information from base station broadcasts,it may broadcast this trust to its neighbors,so that neighbors can update the trust information about that node.


If you can't follow anything that I have written (coz of my poor command over English) please ask me to clarify..

I want to know the problems / loopholes associated with this  framework prior to developing.



Hoping a reply from you .

Thanks !!

[Tommaso Pecorella]

Hi,

On Sunday, July 20, 2014 11:37:34 AM UTC+2, silsila...@gmail.com wrote:

Following is a routing framework proposed for my thesis work. Could you please check it and identify the problems associated when developing such a protocol??

link please... :)

 

I have taken Aodv as the base protocol.Since Aodv is not trusted i want to include trust. For that, in addition to routing able , another "Neighborhood table (NT)" is proposed wherein the neighbours of a particular node as well as trust value associaed with them will be stored.

[...]

Your English is ok, but I'd have to check the whole idea more carefully to know if there are big loopholes.

The one I can see right now is the BS role. There are two things:

1) The BS will be able to broadcast info to its 1-hop nodes. The 2-hop nodes will need to rely on other things.

2) You will need to assume that the nodes can trust the BS, and that an attacker can not become the BS.

I think that you first have to draw a scenario, and then decide if you want to have a BS or not. AODV is usually used in ad-hoc mode, and there's no BS in ad-hoc mode.

The second problem is the scenario AND the reference radio model. You mentioned a WSN, but (despite the fuss that some companies do) Wi-Fi is not a good candidate for WSNs. 

The radio model is important in order to understand if you can use some of the MAC-level security elements.

The third thing to understand is the type of attack you want to focus on.

- Your nodes are already in a (trusted) network and an attacker is threatening them ?

- Your nodes are all in a (trusted) network but you want to isolate some misbehaving nodes ?

- You want to avoid an attacker to enter in your network ? (Authentication)

- You want to avoid to join a malicious network (Rogue network) ?

What kind of security you are considering ? Probabilistic or strict ?

A lot of questions, ins't it ?

Anyway, if you send me the thesis proposal I'll be happy to give it a look.

Cheers,

T.

[silsila...@gmail.com]

Thank you for your detailed reply.

I actually posted a reply to "Reply Privately to author" in ns3 groups in the post "Is aodv a wireless sensor network protocol?", but I don't know what happened. :(

I took some time to reply because I was searching answers for your questions and understanding what you have said.

Please note letters in blue correspond to your words.

" The one I can see right now is the BS role. There are two things:

1) The BS will be able to broadcast info to its 1-hop nodes. The 2-hop nodes will need to rely on other things.

2) You will need to assume that the nodes can trust the BS, and that an attacker can not become the BS. 

I think that you first have to draw a scenario, and then decide if you want to have a BS or not. AODV is usually used in ad-hoc mode, and there's no BS in ad-hoc mode."

By Base Station , I meant some powerful node capable of high end processing and high bandwidth. So the signals sent from this node will be capable of reaching the entire network (Assuming not so large network) . Or if it can't,there may be a boundary up to which the signals may reach and from there the nodes there will be rebroadcasting the signals. If the name Base station does not apply to such a node ,I may change the name.
Nodes trust BS.I assume that BS can't be compromised.

Scenario
*********

BS broadcasts a message which contains info about undelivered packets in a network.

Nodes analyse this information and calculate trust value of neighbours based on this. 

The second problem is the scenario AND the reference radio model. You mentioned a WSN, but (despite the fuss that some companies do) Wi-Fi is not a good candidate for WSNs. 

The radio model is important in order to understand if you can use some of the MAC-level security elements.

I use Two ray ground propagation radio model and mac 802.11 as the mac type. Do this scenario create any problem? If yes,please suggest the appropriate radio model.

The third thing to understand is the type of attack you want to focus on.

- Your nodes are already in a (trusted) network and an attacker is threatening them ? 

Yes.Since Aodv is trusted. There are malicious nodes dropping or replaying packets.

- Your nodes are all in a (trusted) network but you want to isolate some misbehaving nodes ?

Yes.I want to isolate the misbehaving node by decreasing the trust value associaed with them.

- You want to avoid an attacker to enter in your network ? (Authentication)

I do not plan authentication right now

- You want to avoid to join a malicious network (Rogue network) ?

Not planned right now

I want to implement blackhole,wormhole,sinkhole attacks in the network. Since the network is not trusted at first, attacks will be there. Now these should be detected and prevented. The proposed framework is assumed to detect the malicious nodes and isolate them. For that trust value associated with them will be decreased.So in the next run, nodes will not be selecting the malicious nodes to forward the packets.

Awaiting your reply

Thanks for helping me !!!!

[Tommaso Pecorella]

Hi,

thanks for the model clarification. Now it's clear where the issue is.

Let's analyze the "BS" capabilities. You said it is powerful enough to reach all the nodes (and I assume that the nodes can't do that). Is this possible ? Maybe...

In a Tx/Rx, there are at least two elements to be considered: the Tx power and the antenna(s) gain. The antenna gain is involved in Tx *and* Rx (one is receiving!).

If the "BS" wants to be heard by all nodes, it can increase its Tx power, but if it wants to hear all nodes, then all the nodes will have to use the same high Tx power *or* the BS must have an incredibly efficient antenna.

Ok, let's suppose that the BS does have the incredibly efficient antenna. Why the nodes can't have it as well (and save a lot of Tx power?). Usually the answer is: the incredibly efficient antenna is physically larger.

Then, we're in a case where the nodes have a small omnidirectional antenna and the BS is equipped with powerful directional antennas, much like in cellular systems.

Still, can the BS reach ALL the nodes ? What is the needed BS antenna gain ?

Some math is here: http://www.moxa.com/newsletter/connection/2008/03/Figure_out_transmission_distance_from_wireless_device_specs.htm

Mind that for 2.4GHz band, however, walls and vegetation reduce the radio range a lot. As a consequence, the numbers are valid only for a perfectly "clean" Line of Sight.

Summarizing: I fear that the assumption of a BS able to reach all the nodes is a bit strong. Moreover, the interference generated by the BS (when it Tx something, all the nodes are affected) will cause some issues.

If you're totally fine with this, then you can call it "BS", but I'd name it "Master Mode" or something similar, just to avoid confusion.

The last (and perhaps most problematic) question, however, is... why do you need to use an ad-hoc, multihop network if you have the Master Node ?

You're in a cellular-like situation. All the nodes can communicate directly with a node they trust. Every node is 2-hop away and the intermediate is a trusted device.

Why should they want to communicate directly ?

- To save energy ? Nope, because they have to make sure that the master node can "hear" them communicating, so they can't reduce their Tx power.

- To save bandwidth ? Maybe, it's the same scenario of LTE D2D, but the saving only applies in extremely specific cases.

Most of the time, the Master Node mediated communication is more efficient (and more trustable).

And this is the issue... if you have a "BS", you don't need multi-hop :(

Sorry for the rather pessimistic view.

Cheers,

T. 

[Tommaso Pecorella]

Hi,

in the paper you mention the authors did not assume that the BS can "hear" all the nodes.

The authors did assume that the network is a pure gathering system, i.e., that all the nodes wants to send the data only to the BS.

Your scenario is completely different. The trust management system proposed in the paper can not be applied as it is.

Sorry,

T.

On Thursday, July 31, 2014 11:34:01 AM UTC+2, silsila...@gmail.com wrote:

Thanks again

I understand the risks.

"If the "BS" wants to be heard by all nodes, it can increase its Tx power, but if it wants to hear all nodes, then all the nodes will have to use the same high Tx power *or* the BS must have an incredibly efficient antenna.  

Ok, let's suppose that the BS does have the incredibly efficient antenna. Why the nodes can't have it as well (and save a lot of Tx power?). Usually the answer is: the incredibly efficient antenna is physically larger.

Then, we're in a case where the nodes have a small omnidirectional antenna and the BS is equipped with powerful directional antennas, much like in cellular systems."

OK I am not mentioning BS any more. Because I understand that the name "BS" is creating a lots of confusion. As You said ,I name it as a Master node.

One thing is that I do not want master to hear all nodes with one hop.  Master gets packets exactly like all other nodes receive. It gets packets forwarded from other nodes. I have already said Master is a node to where most of the communication goes. For example ,it may be deriving data from the nodes in the network to process them and calculate some results. Nodes do not need anything to equip themselves to send data to Master. 

If this situation also creates any problem,leave every special capability assigned to Master. So if the Master needs to broadcast, the normal broadcasting rules may be applied.

Its not like a cellular system. In cellular systems,every call need to reach BS.But here it is not the case. If a node wants to send anything,it may follow Aodv protocol to deliver its packets. 

"The last (and perhaps most problematic) question, however, is... why do you need to use an ad-hoc, multihop network if you have the Master Node ?

You're in a cellular-like situation. All the nodes can communicate directly with a node they trust. Every node is 2-hop away and the intermediate is a trusted device.

Why should they want to communicate directly ?"

I assume that You got answer for this question. I am not in a cellular like station. There are large number of nodes deployed in a network. Some of them may be malicious. Nodes communicate with each other. I introduced concept of Master because I need information to isolate the malicious ones.

A routing framework proposed for the WSN is as follows.I need to introduce some modifications into this and implement this framework in Aodv since Aodv is not at all trusted.

http://www.cs.wayne.edu/guoxing/contentFrame/Publications/Paper/EWSN10.pdf

refer page 70,71.. , block diagram in page 72.

Hoping another reply wih new issues. 

Thanks :)