There are probably many options to explore in such cases. For example, if the sensors are to be mostly in sleep mode, to save energy, then you can postulate that the sensors will want to wake up periodically, to transmit their data, and go back to sleep. For that approach, you could use something similar to the original Ethernet CSMA/CD algorithm, where collisions result in a random backoff and retry.
That assumes there are many sensors, and that collisions are likely. If there are few sensors, and collisions unlikely, you may be just as happy to let collision occur occasionally, and just wait for the next scheduled update. The easiest way to set up a scheme like this is to introduce some randomness in the sensor message period, or alternatively, to time the next update based on the last SUCCESSFUL transmission, for that particular sensor. You want to avoid having all sensors using exactly the same update schedule, and synchronizing their schedule to collide every time.
Alternatively, the power-starved sensors may possibly have a power-efficient monitoring mode, where they sit mostly idle, waiting to be polled. Only when polled by the base station do they take their measurement, and transmit the data message. With a polling scheme, you will have no collisions. The disadvantage of this scheme is that the sensor has to be monitoring for a poll. But, to mitigate this, you can schedule that too! Begin monitoring for a poll only at a predetermined interval, based on the last successful poll.
So, many strategies are possible. Much depends on system requirements and sensor design. The very simplest method is the one where collisions don't matter and an occasional missed update doesn't matter.
Dear Debabrata
There are many strategies to reduce the collitions, but some question have to be answer firts, such as, what is your MAC protocol? and if you area able to change it?.
Even if you can not introduce change in your MAC, there area some strategy that you can follow at the level of application layer. You can introduce some random waiting time to space the send of messages, or you can add or concat small message in a bigger one to reduce the number of messages and also the probability of collition, but any strategy have to be evaluated because deppending on your network traffic condition the effect could be very different.
To Reduce the power of the signal can reduce its range but it also decrese the number of effective node to collide with.
But a interesting question is, it is really matter the number of collision?, or what is really importan is the effective delivery of the messages?.
Even than to reduce the number of collision seems to be good, because a collision means a waste of time and energy, just a lower number of collision is good?, even is the meesage do not arrive to its final destination?
Let me know if you are interested in additional information.
Cognitve radio algorithm is the best technology for minimizing collison.The spectrum sensing done by cr devices sense the spectrum first and if they find it empty then only they make the entry of secondary device or user.But the algorithm has to be developed efficiently .The algorithm can be coupled with csma cd technology for minimizing collison as well as interference
There are probably many options to explore in such cases. For example, if the sensors are to be mostly in sleep mode, to save energy, then you can postulate that the sensors will want to wake up periodically, to transmit their data, and go back to sleep. For that approach, you could use something similar to the original Ethernet CSMA/CD algorithm, where collisions result in a random backoff and retry.
That assumes there are many sensors, and that collisions are likely. If there are few sensors, and collisions unlikely, you may be just as happy to let collision occur occasionally, and just wait for the next scheduled update. The easiest way to set up a scheme like this is to introduce some randomness in the sensor message period, or alternatively, to time the next update based on the last SUCCESSFUL transmission, for that particular sensor. You want to avoid having all sensors using exactly the same update schedule, and synchronizing their schedule to collide every time.
Alternatively, the power-starved sensors may possibly have a power-efficient monitoring mode, where they sit mostly idle, waiting to be polled. Only when polled by the base station do they take their measurement, and transmit the data message. With a polling scheme, you will have no collisions. The disadvantage of this scheme is that the sensor has to be monitoring for a poll. But, to mitigate this, you can schedule that too! Begin monitoring for a poll only at a predetermined interval, based on the last successful poll.
So, many strategies are possible. Much depends on system requirements and sensor design. The very simplest method is the one where collisions don't matter and an occasional missed update doesn't matter.
Thanking you .
Proff. Ricardo González , Proff. Mehboob Ul Amin , Proff. Albert Manfredi
Proff. Timothy Kreps, I am following your paper for minimization Of collision in WSN.
- Badges
- Science topic