Performance Analysis

allows a downlink packing algorithm to achieve a high utilization of the resources is presented.

Sorting the MSs resource request in decreasing order before packing will definitely impact on how efficient the resource in the downlink subframe is going to be used by the packing algorithm because is more easier to fit small burst into the remaining spaces on the subframe, specially when this is the last remaining burst that need to be packed and on the contrary, a larger burst not packed will lead to more unused slots.

But, special care has to be taken when designing a packing algorithm since operations like ordering the data to be packed by decreasing size might result in a larger utilization of the WiMAX downlink subframe resources but not necessarily in a larger throughput. MSs with better channel conditions will have a better MCS and thus, one slot can transmit more data than one MS with lower MCS.

So, in order to achieve higher throughput and an efficient use of the resources in the downlink subframe, it is advised to pack based on the amount of slots the MS require and the total data that would be packed. For example, pack the burst with the highest amount of data to transmit first, in case of a tie, refer to the amount of slots assigned, selecting the highest. Consider the following scenario, a small burst with higher MCS and a large burst with lower MCS and less data to transmit, in this case is better to pack the smaller in order to achieve higher throughput but in some cases this might leave more unused spaces as said before.

Obviously, the reason why the proposed algorithm performs better than eOCSA and OBBP is the way of allocating the burst into the downlink subframe. Because bursts are required to have a rectangular shape, the proper dimension of it is another procedure that needs our attention while designing an efficient packing algorithm.

A burst can have different dimensions with different resulting over allocation slots or no over allocation at all. Then, selecting the dimension with none or less over allocation should be the best in order to maximize the utilization of resources in the downlink subframe. It‟s believed that packing efficiency of the proposed algorithm could increase a little more if applied, but was not considered to lower the complexity of the algorithm. Instead, like in the eOCSA algorithm, most of the bursts allocated by the algorithm have the least widths, which imply that the active time and consequently, the energy consumption of each MS is minimized.

While allocating bursts vertically or column wise simplifies things, it may results in some disadvantages. The problem of packing burst column wise or row wise is that, for example, in eOCSA all bursts in a common column will have the same width, and so, the burst will only adopt one dimension. For this reason, some burst may result in more over allocated slots than others. Besides, the whole column can possibly not be filled completely with burst, given the reason that there‟s no resource request that can fill the spaces left. And this will apply for all the columns of burst, resulting into more unused spaces, as shown on figure 15. Notice that, only the first column of burst, composed by request A₄ and A₁₀ was completely filled in this example.

Figure 15: Illustration of the wasted slots of eOCSA algorithm

If we forget about how eOCSA compute its burst dimension, burst A₅ dimension (on figure 15) could have been set with height equal to 4 and width equal to 2, resulting in only one over allocation slot, and if we didn‟t have this column width limits, the space left could have been used to place other burst more flexibly. Figure 16 illustrates the above mentioned; where the white area is the unallocated slots and the over allocated slots are shown in black.

Figure 16: Illustration of wasted space produced by mapping burst column wise

Like in eOCSA algorithm, OBBP can also have unused slots on the top of each columns of burst. OBBP sort their columns in descending order, so that all unallocated spaces will be located to the left and top of the downlink subframe, please refer back to figure 7 on chapter 3. By doing this, all unused slots will be together, making it easier to place a burst that could possibly not fit without this. Then, they try to allocate the burst in a selected rectangle that result with the least over allocated slots, but because they still continue with the use of their OFs, the dimension of some burst left cannot be placed into some of these rectangles, which is the reason why OBBP results in more unused slots than the compared schemes.

Chapter 6:

Conclusion

This work presents a novel downlink burst allocation algorithm for IEEE 802.16e Mobile WiMAX networks. Similar to eOCSA, the proposed algorithm meets the rectangle shape allocation constraint, achieves high throughput by considering mapping the larger resources first, and optimizes energy consumption at MS by minimizing its receive time. The basic idea of our proposed algorithm is to find the maximum rectangle in the un-allocated space for allocation in each iteration. An efficient method for finding the maximum rectangle is provided. Furthermore, the algorithm is enhanced to improve the efficiency. The performance of the proposed algorithm is compared with that of eOCSA and OBBP. Simulation results show that the proposed algorithm, enhanced maximum rectangle-based burst allocation algorithm, outperforms eOCSA and OBBP.

Mapping bursts column wise or row wise can reduce the complexity of packing algorithms but may not result in the best efficient way of packing. A packing algorithm that strictly tries to maximize the utilization of the frame can result in a violation of the agreed QoS guarantees or unfair distribution of the resources.

Therefore, a trade-off between radio resource usage maximization, QoS and fairness needs to be considered.

References

[1] Byeong Gi Lee and Sunghyun Choi, “Broadband Wireless Access and Local Networks: Mobile WiMAX and WiFi”, Artech House, 2008.

[2] T. Ohseki, M. Morita, and T. Inoue, “Burst Construction and Packet Mapping Scheme for OFDMA Downlinks in IEEE 802.16 Systems,” Proceedings of IEEE Global Comunications Conference (IEEE GLOBECOM), 2007.

[3] Y. Ben-Shimol, I. Kitroser, and Y. Dinitz, “Two-dimensional mapping for wireless OFDMA systems,” in IEEE Trans. Broadcasting, vol. 52, no. 3, pp.

388-396, Sept. 2006.

[4] Bacioccola, C. Cicconetti, L. Lenzini, E.A.M.E. Mingozzi, and A.A.E.A. Erta,

“A downlink data region allocation algorithm for IEEE 802.16e OFDMA,” in Proc. Int. Conf. Information, Comm. and Signal Processing, 2007, pp. 1-5.

[5] F. Clautiaux, J. Carlier, and A. Moukrim, “A new exact method for thetwo-dimensional orthogonal packing problem,” in European Journal Operational Research, vol. 127, pp. 1196-1121, Dec. 2007.

[6] Patrick Hosein, “Cross-Layer Design for Data Burst Construction in the Downlink of IEEE 802.16 Systems”, Proceedings of IEEE Global Comunications Conference (IEEE GLOBECOM), 2008.

[7] X. Perez-Costa, P. Favaro, A. Zubow, D. Camps and J. Arauz, “On the Challenges for the Maximization of Radio Resources Usage in WiMAX

Networks”, in Proc. IEEE Consumer Communications and Networking Conference (CCNC 2008), pp. 890-896, 2008.

[8] So-In, R. Jain, and A. Al-Tamimi, “OCSA: An algorithm for Burst Mapping in IEEE 802.16e Mobile WiMAX Networks”, Proceedings of the 15th Asia-Pacific Conference on Communications. (APCC 2009), Oct. 2009.

[9] So-In, R. Jain, and A. Al-Tamimi, “eOCSA: An Algorithm for Burst Mapping with Strict QoS Requirements in IEEE 802.16e Mobile WiMAX Networks”, Proceedings of the 2nd International Conference on Computer and Automation Engineering (ICCAE 2010), Feb. 2010.

[10] Eshanta, M. Ismail, K. Jumari, “An efficient burst packing algorithm forOFDMA systems”, Proceedings of the 2nd International Conference on Computer Technology and Development (ICCTD), Nov. 2010.

[11] K.-P. Shih, C.-T. Chiang, T.-H. Hsieh, and Y.-H. Lin, “A downlink resource management scheme in IEEE 802.16 OFDMA-based system”, Proceedings of the Workshop on Wireless, Ad Hoc, and Sensor Networks, Sep. 2008.

[12] C. Cicconetti, L. Lenzini, A. Lodi, S. Martello, E.Mingozzi, M. Monaci,

“Efficient Two-dimensional Data Allocation in IEEE 802.16 OFDMA”, Proceedings of IEEE IEEE International Conference on Computer Communications (IEEE INFOCOM), 2010.

[13] Panagiotis G. Sarigiannidis, Malamati D. Louta, Periklis Chatzimisios, “A Novel Fair Mapping Scheme for IEEE 802.16Downlink Sub-Frame”, Proceedings in International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT), 2010.

Biography

Jimmy Yau Zhong was born in July1, 1985 in Panama City, Panama. Jimmy entered Technological University of Panama as an undergraduate in 2004 after completing his elementary and high school studies at Saint George International School of Panama. In 2008, he completed his Bachelor of Computer Networks. Later that same year, he moved to Taiwan to start his graduate studies at National Chiao Tung University.