|
|
WiFi-Nano
The WiFi capacity has significantly increased, from 1 Mbps to 1 Gbps over the last few years. However, the maximum achievable throughput did not follow, leading to an efficiency as low as 10 % at 1 Gbps.
Fig. 1. WiFi-Nano transmission at a data rate of 600 Mbps [1].
To overcome this issue, Microsoft Research recently created novel WiFi PHY/MAC protocols referred to as WiFi-Nano, which protocol is depicted in Fig. 1. Efficiency gains as high as 100 % were experimentally demonstrated using FPGA-based software defined platforms.
Motivation
Fig. 2. Overhead components in IEEE 802.11 at 600 Mbps [1].
The main MAC layer overheads come from the channel access, data preamble, acknowledgement (ACK), and collision, as depicted in Fig. 2. Fig. 2 depicts these overheads assuming a transmission of a given frame with a typical payload length of 1500 bytes at a data rate of 600 Mbps. As one can see, the channel access overhead is major, representing 500 % of the air time compared to the payload transmission. Prior to the transmission of data, the preamble component is crucial to synchronize and help in channel estimation. The ACK overhead is defined as the switching time from reception mode to transmission mode and takes also into account the transmission of the ACK frame to notify the sender about the proper frame reception.
Fig. 3. Fraction of air time at different data rates [1].
By varying the data rate we can compare the fraction of air time for the overheads and data transmission, as depicted in Fig. 3. As we can observe, as the data rate increases the fraction of air time of data transmissions decreases. That is, the overhead components significantly grow as the data rate increases. Given these observations, there is a clear need for reducing MAC overheads to improve WiFi efficiency.
WiFi-Nano Overview
Fig. 1 depicts the approach used in WiFi-Nano, whose overall process is much simpler. When a given node is ready to transmit data, after the binary exponential backoff (BEB) mechanism, the preamble is sent using the first antenna and the second one listens to preambles from other stations. Fig. 4 presents a more detailed execution of this process, referred to as speculative preambles.
Fig. 4. Speculative preambles [1].
The speculative preambles process enables carrier sensing while transmitting and channel contention is performed simultaneously during preamble transmissions. In the above example, node B aborts since it detects that node A started its preamble first. Note that it takes 5 slots by the lattice correlator in order to detect the preamble, a mechanism described below and depicted in Fig. 5.
Fig. 5. Lattice correlator used to interpret receiving preambles [1].
The lattice correlator provides two functions. First, it correlates the sub-parts of a given preamble. The preamble begins with a set of pseudo-random noise (PN) sequences (PN1, PN2, ..., PN5), whereby each PN consists of a set of samples. Then, executing the additions in Fig. 5 allows to detect the preamble with high probability, as shown in the next section. The second function is the detection of the transmission start time of the first preamble, which is used to roll back the backoff counter properly.
Experimental and Simulation Results
As mentioned in the previous section, correlating sub-parts of the preamble is crucial in WiFi-Nano. In order to improve detections, the self-interference cancellation technique has been used to remove the transmitted signals from the received signals of other stations.
Fig. 6. Preamble detection in WiFi-Nano [1].
Fig. 6 presents experimental results of the lattice correlation technique with and without self-interference cancellations. With the self-interference cancellation technique, preambles were detected even at an SNR equal to 0. If self-interferences are not removed, a better signal quality (SNR) is required.
Fig. 7. Throughput performance of WiFi-Nano compared to legacy WiFi based on IEEE 802.11 [1].
Fig. 7 shows the throughput performance of WiFi-Nano compared to legacy WiFi based on IEEE 802.11. The obtained results are show significant performance improvements, whereby the throughput gain of WiFi-Nano over IEEE 802.11 is as high as 100 % at a data rate of 600 Mbps.
Reference
[1]
|
E. Magistretti, K. K. Chintalapudi, R. Ramjee, and B. Radunovic,
“WiFi-Nano: Reclaiming WiFi Efficiency Through 800 ns Slots,”
Proc., ACM International Conference on Mobile Computing and Networking (MobiCom), pp. 37-48, Sept. 2011, Las Vegas, NV, USA.
|
|
|
|
| |