In this section, simulations are conducted in order to evaluate the performance of the proposed PSWD approach. It is also in comparison with the behavior of sleep mode operations for IEEE 802.16e and 802.16m in aspect of either Type I or Type II. A single BS/MS (ABS/AMS) pair is considered as the simulation scenario. All the required negotiations and parameter update procedures among both ABS and AMS for the
sleep mode operation implemented via an MATLAB event-driven simulator. It is noted that the traffic state estimation process, calculation of evaluation metrics, and the sleep window determination policy of the proposed PSWD approach are all executed only at the ABS side and then inform the affiliated AMS about the next action to be undertaken, since the ABS has no power-saving concern. Consequently, the AMS should also provide its own UL traffic information observed during the previous sleep window to the serving ABS via control messages originally defined in the IEEE 802.16m standards. The parameters adopted in the simulations are list in Table I wherein the values of energy consumption are in light of the industrial manufactured mobile WiMAX chip [30].
TABLE I : SIMULATION PARAMETERS
Parameter Value
Frame duration 5ms
Idle period (τ ) 4 frames
Initial sleep window/cycle length of Type I for 16e/16m 1 frame Maximum sleep window/cycle length of Type I for 16e/16m 128 frames Overall sleep window length of Type II for 16e&16m 4 frame
Default listening window length 1 frame
Energy consumption per busy frame (εB) 280 mW
Energy consumption per idle frame 120 mW
Energy consumption per sleep frame (εS) 10 mW Energy consumption per state switching (εSW) 1 mW
Mean service rate (µ) 3 packets/frame
Each simulation run is carried out for 10 minutes, and every acquired outcome with the individual mean arrival rate is averaged from 100 simulation runs. Furthermore, it is noted that a 3-step value function is utilized in the PSWD approach, where the
1.1 1.105 1.11 1.115 1.12 1.125 1.13
1.1 1.105 1.11 1.115 1.12 1.125 1.13
x 104 0
1
Num. of Pkts
UL Traffic State (λu = 0.02)
1.1 1.105 1.11 1.115 1.12 1.125 1.13
x 104 0
1
MS State
IEEE 802.16e Type I
1.1 1.105 1.11 1.115 1.12 1.125 1.13
x 104 0
1
MS State
IEEE 802.16e Type II
1.1 1.105 1.11 1.115 1.12 1.125 1.13
x 104 0
1
AMS State
IEEE 802.16m Type I
1.1 1.105 1.11 1.115 1.12 1.125 1.13
x 104 0
1
AMS State
IEEE 802.16m Type II
1.1 1.105 1.11 1.115 1.12 1.125 1.13
x 104
Figure 6.3: An exemplified sleep mode operation among Type I/II of IEEE 802.16e/m and PSWD approach.
discount factor γ for the future ith step value function is chosen as (0.5)i, i = 1, 2, for the reason that the rewards in later time are considered less influential to the present action.
Fig. 6.3 depicts a exemplified timing diagram of sleep mode operation among all above schemes, including the IEEE 802.16e, the IEEE 802.16m, and the proposed PSWD approach under packet arrival rates of λd = 0.1 and λu = 0.02. The two sub-plots from the top show the number of arriving packets from DL and UL directions, respectively, within the frame duration [11000, 11300] of the simulation. The corre-sponding operations of each scenario are presented in the remaining subplots below, where it is noticed that the state “1” represents that an MS/AMS is staying in the normal mode or in listening windows; while the status of power-saving, i.e. sleep state
0.1 0.2 0.3 0.4 0.5
Figure 6.4: Performance comparison among Type I of IEEE 802.16e/m and PSWD approach under NRT/BE traffic.
of an MS/AMS is denoted as state of “0”. As can be observed from Fig. 6.3, the characteristics of individual scheme is obviously performed, e.g. Type I of 16e/m is suitable for non real-time and BE traffic, on the other hand, Type II is preferred for real-time traffic-only or real-time and BE-traffic mixed scenarios thanks to its periodic traits. The 16e MS of either Type I or Type II tends to spend more time duration in the awake state than the 16m AMS due to the defect of existence of the normal mode in 16e. Nevertheless, compared to the proposed PSWD approach, both 16e and 16m seem to be inefficient on account of numerous under-utilized listening windows caused by adopting the binary-exponential growing mechanism. The AMS of PSWD almost keeps staying in the sleep state (the awake states of listening widows are at frame number 11000, 11129 − 11130, and 11259 − 11264, and remainders are resulted from the interruptions due to UL transmission), since each sleep window is determined by rewards of the POMDP model according to the estimated present traffic state.
Fig. 6.4 shows performance comparisons between 16e, 16m, and the proposed PSWD approach over various arrival rates of DL traffic with loose delay constraints, such as BE and non-real-time traffic. Each λd is an average value from the dMMPP traffic model, and the arrival rate of UL traffic is fixed at λu = 0.02 packets /frame. Note that as for these traffic types, the performance of mean packet delay is not of great importance, so that it can be turned to achieve higher power-saving efficiency. Due to the Reward Assignment Algorithm in PSWD, the outcomes of mean packet delay satisfy the respective tolerable delay δ 1, as for the case that does not consider any delay constraints, the results are bounded by half the maximum size of sleep window, i.e. the remaining vacation time which equals to 128/2 = 64 frames. As can be seen from the metric of energy cost, it is expected that the average energy consumption increases with augmented arrival rates of DL traffic. Moreover, the proposed PSWD can acquire the relatively low energy consumption in comparison with the conventional 16e/m scheme, this may be attributed to the selected length sleep window is more accurate than binary-exponential algorithm, which always attempts to fit the traffic from the minimum size of initial sleep window.
On the other hand, the real-time traffic and BE-traffic mixed scenarios with demand of stringent delay constraint are presented in Fig. 6.5, wherein the corresponding power-saving class of Type II are adopted for both 16e and 16m. With similar values of packet delay, the proposed PSWD (δ = 20 ms) can provide better performance with regard to the average energy cost when the arrival rate is augmented (λd > 0.5). It is also observed that each outcome of packet delay for the PSWD approach meets the values of tolerable delay.
Fig. 6.6 tries to investigate the influence of the ratio of DL and UL traffic, for it is of interest that user behavior in the Internet through the IEEE 802.16 WiMAX
1Note that the y-axis of the figure about mean packet delay is in unit of frame; while the delay constraint δ comes in ms.
0.1 0.2 0.3 0.4 0.5
Figure 6.5: Performance comparison among Type I of IEEE 802.16e/m and PSWD approach under RT-only or RT and BE mixed traffic.
infinity 10:1 5:1 2:1 1:1 20
infinity0 10:1 5:1 2:1 1:1 5
Figure 6.6: Performance comparison among Type I of IEEE 802.16e/m and PSWD approach under different DL/UL ratios.
networks. The x-axis of the figure shows various values of λd : λu ratio, where “infinity”
corresponds to the situation with only DL traffic. The ratio of “10 : 1” may occur in the case that the user is surfing the Internet by a handhold mobile phone; while a regular user using a computer for watching a video stream via P2P technique may introduce the same amount of traffic between DL and UL (1 : 1). As shown in Fig. 6.6, the more energy cost is required when the proportion of UL traffic becomes higher, since the MS/AMS needs additional duration of time for transmitting the UL packets. The situation is particularly evident as for the case of 16e, wherein the sleep mode of the MS would be interrupted for the occurrence of UL traffic and switched back to the normal mode immediately. On the other hand, the mean packet delay tends to decrease when the DL/UL ratio is augmented due to the reason that during the UL transmission, both DL and UL data transportation are allowed at the same time via the super frame with individual DL-MAP and UL-MAP. It is found that the PSWD with the large delay constraint (δ = 160 ms) gains noticeable improvement since the remaining vacation time has been diminished by half, which turns to be a great quantity in packet delay related to other schemes.
Chapter 7 Conclusions
In this thesis, two comprehensive analytical system models are proposed according to the sleep mode operations of Type I/Type II for the IEEE 802.16e and the IEEE 802.16m broadband wireless networks respectively. Both the downlink and uplink traf-fic are considered simultaneously, with sleep ratio and mean packet delay as the mea-sures for performance analysis. The effectiveness of the analytical models are validated through numerical simulations. Furthermore, a POMDP-based sleep window determi-nation (PSWD) approach for improving the performance of the sleep mode operation of the IEEE 802.16m is presented. The PSWD approach resolves the length of each sleep window based on the rewards calculated by means of a POMDP model, which spec-ulates the present traffic state via the concept of belief states at each decision epoch.
The efficiency of the PSWD approach is evaluated by simulations in terms of energy cost and mean packet delay with tolerable delay taken into account. Simulations show that the proposed PSWD approach outperforms the conventional IEEE 802.16e and IEEE 802.16m corresponding to various traffic demands and satisfies respective delay constraints at the same time.
Bibliography
[1] IEEE Standard for Local and Metropolitan Area Networks - Part 16: Air Interference for Fixed Broadband Wireless Access Systems, IEEE Standard 802.16-2004, June 2004.
[2] IEEE Standard for Local and Metropolitan Area Networks - Part 16: Air Interference for Fixed Broadband Wireless Access Systems, Amendment 2: Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands and Corrigendum 1, IEEE Standard 802.16e-2005, Dec. 2006.
[3] IEEE Standard for Local and Metropolitan Area Networks - Part 16: Air Interference for Fixed Broadband Wireless Access Systems, IEEE Standard 802.16-2009, May 2009.
[4] DRAFT Amendment to IEEE Standard for Local and Metropolitan Area Networks - Part 16:
Air Interference for Fixed Broadband Wireless Access Systems – Advanced Air Interface, IEEE P802.16m/D7, July 2010.
[5] M.-G. Kim, J. Choi, and M. Kang, “Scheduled Power-Saving Mechanism to Minimize Energy Consumption in IEEE 802.16e Systems,” vol. 12, no. 12, pp. 874–876, 2008.
[6] T.-C. Chen and J.-C. Chen, “Extended Maximizing Unavailability Interval (eMUI): Maximizing Energy Saving in IEEE 802.16e for Mixing Type I and Type II PSCs,” vol. 13, no. 2, pp. 151–153, 2009.
[7] Y. Xiao, “Energy saving mechanism in the IEEE 802.16e wireless MAN,” IEEE Communications Letters, vol. 9, pp. 595–597, 2005.
[8] Y. Zhang, “Performance Modeling of Energy Management Mechanism in IEEE 802.16e Mobile WiMAX,” Proc. IEEE WCNC, pp. 3205–3209, 2007.
[9] J. Almhana, Z. Liu, C. Li, and R. McGorman, “Traffic Estimation and Power Saving Mechanism Optimization of IEEE 802.16e Networks ,” Proc. IEEE ICC, pp. 322–326, 2008.
[10] Y. Park and G. U. Hwang, “Performance Modeling and Analysis of the Sleep-Mode in IEEE 802.16e WMAN,” Proc. IEEE VTC, pp. 2801–2806, 2007.
[11] Y. P. Hsu and K. T. Feng, “Performance Modeling of Power Saving Classes with Multiple Con-nections for Broadband Wireless Networks,” Proc. IEEE WCNC, pp. 1477–1482, 2008.
[12] Y. Zhang, Y. Xiao, and V. C. Leung, “Energy Management Analysis and Enhancement in IEEE 802.16e WirelessMAN ,” IEEE Transactions on Vehicular Technology, vol. 58, pp. 3738–3752, 2009.
[13] S. Cho and Y. Kim, “Improving Power Savings by Using Adaptive Initial-Sleep Window in IEEE802.16e,” Proc. IEEE VTC, pp. 1321 –1325, 2007.
[14] M. Peng and W. Wang, “An Adaptive Energy Saving Mechanism in the Wireless Packet Access Network,” Proc. IEEE WCNC, pp. 1536–1540, 2008.
[15] M.-G. Kim, J. Choi, and M. Kang, “Adaptive Power Saving Mechanism Considering the Request Period of Each Initiation of Awakening in the IEEE 802.16e System,” IEEE Communications Letters, vol. 12, no. 2, pp. 106–108, 2008.
[16] A. Anisimov, S. Andreev, and A. Turlikov, “IEEE 802.16m energy-efficient sleep mode operation analysis with mean delay restriction,” International Conference on Ultra Modern Telecommuni-cations and Workshops (ICUMT), pp. 1–4, 2009.
[17] S. Baek, J. J. Son, and B. D. Choi, “Performance Analysis of Sleep Mode Operation for IEEE 802.16m Advanced WMAN,” Proc. IEEE ICC Workshops, pp. 1–4, 2009.
[18] E. Hwang, K. J. Kim, J. J. Son, and B. D. Choi, “The Power-Saving Mechanism With Periodic Traffic Indications in the IEEE 802.16e/m,” IEEE Transactions on Vehicular Technology, vol. 59, no. 1, pp. 319 –334, 2010.
[19] L. Kleinrock, Queueing System, 1st ed. New York : Wiley, 1975.
[20] L. Kong and D. Tsang, “Performance Study of Power Saving Classes of Type I and II in IEEE 802.16e,” Proc. IEEE LCN, pp. 20–27, 2006.
[21] D. Gross and C. M. Harris, Fundamentals of Queueing Theory, 3rd ed. New York : Wiley, 1998.
[22] L. Miller, “A Note on the Busy Period of an M/G/1 Finite Queues,” Operations Research, vol. 23, pp. 1179–1182, 1975.
[23] S. W. Fuhrmann, “A Note on the M/G/1 Queue with Server Vacations,” Operations Research, vol. 32, pp. 1368–1373, 1984.
[24] C. H. Hsu, K. T. Feng, and C. J. Chang, “Statistical Control Approach for Sleep Mode Operations in IEEE 802.16m Systems,” IEEE Transactions on Vehicular Technology, vol. PP, pp. 1–1, 2010.
[25] A. C. Cassandra, “Exact and approximate algorithms for partially observable Markov decision processes,” Ph.D. dissertation, Brown Univ., Providence, RI, May 1998.
[26] L. P. Kaelbling, M. L. Littman, and A. R. Cassandra, “Planning and Acting in Partially Observ-able Stochastic Domains,” Artificial Intelligence, vol. 101, pp. 99–134, 1998.
[27] A. Andersen and B. Nielsen, “A Markovian approach for modeling packet traffic with long-range dependence,” vol. 12, no. 5, pp. 719–732, 1998.
[28] L.Muscariello, M.Mellia, M.Meo, and M. Marsan, “An MMPP-Based Hierarchical Model of In-ternet Traffic,” Proc. IEEE ICC, vol. 4, pp. 2143 – 2147, 2004.
[29] S. Shah-Heydari and T. Le-Ngoc, “MMPP models for multimedia traffic,” Telecommunication Systems, vol. 15, pp. 273–293, 2000.
[30] Sequans-Communications, “Datasheet: SQN1130 System-on-Chip (SoC) for WiMAX Mobile Sta-tions.”