Extension for Achieving Energy Efficiency

Due to the power constraint of sensor nodes, energy efficiency is also an important issue for WSNs. As mentioned above, most CSMA-based MAC protocols can be adopted as the underlying MAC protocol. Below, we will use B-MAC [20] as our choice and show how to utilize its LPL (Low Power Listening) technique to achieve energy efficiency.

In LPL, a sensor normally stays in the sleep state and wakes up periodically.

When a sensor wakes up, it will turn on its radio for a very short duration and check for any activity. If a preamble is detected, the sensor will stay awake to capture the incoming packet. Since nodes are not synchronized and thus wake up at different times, the preambles of data packets should be longer than the check interval of sleeping nodes to ensure that sleeping nodes will not miss incoming packets. More details of LPL can be found in [20].

Below, we make some notes about the combination of our scheme with LPL technique. First, both the TDMA-like protocol and the periodical wake-up scheme need timers. These two timers should be run independently. Besides, the maxi-mum one-way message delay (i.e., d) should include the preamble length. Fig. 6.8 shows an example, where ` = 1. Second, recall that CCA needs to be run before any transmission. If the CCA outlier algorithm observes that the channel is not clear, the sensor should switch to the receive mode instead of going back to sleep.

The reason is that our scheme depends on overhearing for inhibiting reporting and increasing data fusion opportunity.

6.3 Simulation Results

We have developed a simulator to demonstrate the efficiency of our proposed ap-proach. A sensing field with size 256×256 units where 4096 sensors are deployed randomly with uniform distribution is simulated. The sensor with ID 0 is selected to be the sink. In order to simulate the events arising in the network, a simple event generation model is proposed. In this model, we use four parameters to control the generation of events:

• MAX INTERVAL: This parameter defines the maximum time interval be-tween two events.

• WIDTH and MAX LEVEL: In our model, an event area is represented by multiple concentric circles. The number of concentric circles is determined

by MAX LEVEL. The first circle is the one with radius WIDTH, the second circle is the one with radius 2 × WIDTH, and so on.

• PROPAGATION DELAY: This parameter is used to simulate the event propagation delay. When an event occurs, if sensors in the i-th annulus of the event area detect this event at ti, then sensors in the i + 1-th annulus will detect this event at ti+ PROPAGATION DELAY.

Now, we describe the procedure of this event generation model. The first event will be triggered at the beginning of simulation. As we mentioned above, an event area is represented by multiple concentric circles. Therefore, a point in the send-ing field will be selected randomly as the center of those circles. Sensors in the first circle will detect this event first. Then, after PROPAGATION DELAY, sen-sors in the second annulus will also detect this event. This detection procedure will continue until sensors in the MAX LEVEL-th annulus detect this event. Fi-nally, when an event ej arises initially, the next event (i.e., event ej+1) will also be triggered after t, where 0 ≤ t ≤ MAX INTERVAL and t is determined randomly with uniform distribution.

Three metrics are used to evaluate the performance of medium access schemes.

We count the number of packets transmitted. Usually, fewer packets means that sensors can stay in sleep mode longer. Thus, less energy is consumed. We also measure the success rate of packet transmission defined as the ratio of the number of packets received by the intended receiver to the number of packets transmit-ted by the sender. Success rate can be used to evaluate the efficiency of a MAC protocol. Higher success rate means less collision. Average delay is defined as the average delay of report packets received by the sink. Besides, two metrics are used to evaluate the performance of report reduction schemes. Coverage is defined as Acorr reg union/A_{event area}, where Acorr reg union denotes the area of the field united by the correlation regions of reporters whose reports are received by the sink, and A_{event area} is the area of event area. Higher coverage means that the sink has more accurate information regarding events. Finally, for an unit area in

Table 6.1: Parameters used in the simulation for our proposed link-layer protocol.

Buffer Size 10

The length of DATA 30 Bytes

Bit rate 250 kb/s

Simulation Time 1 hour

MAX INTERVAL 10 seconds

WIDTH 10 units

PROPAGATION DELAY 5 milliseconds

MAX LEVEL 5 (Default)

` 1.5 (Default)

the event area, if it is covered by n sensors’ correlation regions, where n > 1, then we define that the redundancy of that unit area is (n − 1) × 100%.

First, we compare our proposed schedule-based approach with a CSMA-based protocol. Also, two report reduction schemes and two slot assignment schemes will be applied separately. The detail will be described later. Then, we further investigate the impact of two parameters used in our proposed approach, that is, α and `. The related parameters used in the simulation are shown in Table 6.1.