The energy consumption of wireless sensors is very critical since often their energy supply comes from a small battery with limited lifetime. Some techniques like STEM (Sparse Topology Energy Management) make use of a low power wake up channel in addition to a high power data channel. The idea is to minimize energy consumption by turning off the data channel when it is not used. However there is a drawback, the low power wake up channel reduces the maximum range. Also, this technique (STEM) has a considerable latency on the wake up process.
This paper proposes a technique named PTW(Pipelined Tone Wakeup) that also incorporates 2 radios, but for the wakeup channel, a tone is transmitted for a period of time long enough for all neighbors to be recognized, so all of them are awakened, but only one of them receives the notification packet, the awakened nodes that do not receive the notification packet will go back to sleep after a timeout expiration. The energy saving is achieved by passing the duty of the wakeup from the receiver to the sender. In this case, the receiver is alternating between sleep and tone monitoring where the listening period is shortened and the sleeping period is increased compared to STEM.
The end to end delay improvement is achieved by overlapping the data transmission time (data channel), to the awakening time (wakeup channel). So during this process, a node will be receiving data and sending the wake up tone at the same time.
This paper deals with a wakeup scheme to achieve balance between energy efficiency and delay. Though the authors claim that the proposed PTW is better than STEM but I think even PTW has issues.In PTW it is not necessary to wake up all its neighbors as it will increase in energy consumption when notification message is lost. I am also not sure how PTW can be applied to multiple events scenario. Graphs need to be explained more clearly.Though this paper is not perfect in every aspect but I can't help liking this paper.
ReplyDeleteI agree, as mentioned in the paper, if a node does not receive any notification packet after being awakened, it will stay active up to a timeout duration, this is a waste of energy consumption.
ReplyDeleteSynchronization is not required in this protocol since the wakeup tone from a node will wake up the neighboring nodes in the transmission range. This is one of the main improvement in this paper. The authors have carefully looked at the trade off between latency and energy efficiency which STEM has failed to handle.
As every one has mentioned, the paper does not address the possibility of a notification message being lost, which may significantly increase power consumption. However, in an ideal scenario, the protocol works well and is light weight in the sense that there is no synchronization involved.
ReplyDeleteWhile it is true that it is not necessary to awake all the neighbor nodes, please take into account that as soon as their internal timer expires, and no message was intended for them, they will go back to sleep. So it is expected to have more energy savings due to the overlap in reception and transmission time than the losses of energy due to innecessary wakeups.
ReplyDeleteThis is a really good technique for a dual radio sensor node, but when designing a network someone will really have to take into account whether or not they can afford up to 15% ($) more per node. This could end up being a relatively large amount of money based on how big the network is going to be.
ReplyDelete