الفهرس 
MM-Wave MAC ChallengesOnly 14 pages are availabe for public view
 |  | from | 122 |  |  |
| | | from | 122 | | |
Abstract
The unlicensed 7 GHz frequency band centered at 60 GHz, commonly known as the millimeter wave (mmW) band, caught the research and industry attention as a possible candidate to extend the usable spectrum especially after the recent success in fabricating cheap mmW transceivers. The purpose of this study is to enhance the throughput of Multi-hop wireless networks with directional links that operates in mm-Wave range. Mm-W has different propagation characteristics than waves in lower frequency bands, like: Blockage and Deafness. Researcher believes that these different characteristics require a change for MAC layer protocols to be able to deliver a sub-optimal data transfer throughput. Basically this research targets the scheduling algorithm of the MAC layer.
Scheduling for simultaneous directional transmissions has been studied in the literature in many works. One of the main proposed algorithms is Greedy Maximal Scheduling (GMS). But some researchers followed a different approach as they tried to reach a TDM- like performance, completely avoided the collisions and also assured that the transmitter and receiver aligned their beams towards each other via a distributed trial and error algorithm called memory-guided directional MAC (MDMAC)
The overhead of sending a packet in MDMAC is very expensive; it takes around 45% of the total packet time, as each packet needs beamforming, Short Interframe Space (SIFS) interval and acknowledgement packet. To avoid such big overhead we are trying to make use of packet aggregation high throughput gain to achieve a higher utilization for the network. This is done by trying to make the slots allocation as contiguous as possible. In other words, scheduling will not only target slot organization to avoid collisions and achieve fairness but will also try to make the best of the air via reserving the same amount of slots for the nodes as contiguously as possible to be able to use packet aggregation and hence, be able to send groups of packets with one beamforming, SIFS and also get one acknowledgement on them in a bitmap-like fashion instead of getting an acknowledgement packet for each sent packet. We propose a scheduling algorithm to achieve that and show that the resulting net throughput under the saturated network model is higher than the schemes that do not take that property into consideration like GMS or MDMAC by 71% on average. Researcher shows that the protocol processing overhead is insignificant. Although our suggested algorithm can be applied in many network topologies but here we are interested in mesh networking because the directional property is very promising in this particular case.
Researcher also tries to further improve the overall network throughput via increasing the number of reserved slots using a simple protocol. The previous block acknowledgement technique was concerned in first place to make the slot contiguous as much as possible to be able to make a use of the block acknowledgement concept here researcher tries to schedule much links concurrently but with fair scheduling and to make the schedule fast adaptive for the traffic changes. It depends on iterations fist it reaches any feasible schedule then tries to reach gradually to better schedules .Researcher showed that that technique can enhance the overall throughput by 6%. It is worth saying that Block acknowledgment technique can enhance the throughput when operates in environments with low level signal to noise ratio. But this iterative method is not affected.
included bibliographical References.