Chapter 3 MPR MAC Protocol with Adaptive Modulation and
3.4 Computer Simulations
In this section, we compare the simulated results of the suboptimal cases and the proposed method mentioned in Section 3.1 and 3.2, respectively. The system deployment is the same as Section 3.1, which is a CDMA network with randomly generated spreading codes. In the first three cases, there are eight users deployed as in Fig. 3-3. The network deployment of the fourth case is given in Fig. 3-13. SNR_max and SNR_min in Table 3-1 represent the SNR level of the nearest and farthest users from the CC, respectively.
Table 3-1 SNR parameters of the four simulated cases
Case 1 2 3 4
SNR_max 20 13 30 28
SNR_min 16 9 26 16
M
8 8 8 24Case 1:
The Joint AMC curve in Fig. 3-9 represents the performance of the proposed method, and the curve labeled as Suboptimal AMC (k) means that the MPR matrix is
formed based on the parameters of mode k and the AMC is directly incorporated into MPR environment through the flow described in Fig. 3-2. It can be observed that the Joint AMC curve outperforms the other two suboptimal cases in the region p>0.45 (p:
packet generating probability). The Joint AMC curve also attains the channel capacity of the MPR environment incorporated with AMC, which means that through Joint AMC-MAC design, the MPR capabilities of the physical layer can be fully exploited.
In addition, it can be observed in Fig. 3-10 that although the Suboptimal AMC (1) curve outperforms Joint AMC in the delay performance, its throughput performance is worse than the Joint AMC curve.
It is observed that there exist some losses of the Joint AMC curve compared with the Suboptimal AMC (2) curve in the low traffic region (p<0.45) in Fig. 3-9. The reason of the low traffic loss is that in previous MPR works, it is assumed that all the selected n0 users are active to transmit packets, i.e. full load assumption. But this assumption is invalid in the low traffic region because the selected users might be idle.
Thus in the proposed method, the invalid assumption causes that the interference levels are overestimated for all selected users, therefore CC assigns some weaker users Mode 1 to combat the overestimated interferences. In fact, those weak users could be assigned Mode 2 since the actual interference levels might be low in the low traffic region. Therefore, since users’ default modes are set to Mode 2 in Suboptimal AMC (2), it has more information bits in a packet and achieves a higher effective throughput than Joint AMC in the low traffic region.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fig. 3-9 Throughput comparison between Joint AMC and suboptimal methods for Case 1 (SNR_max = 20 dB, SNR_min = 16 dB)
Case 2:
Simulation in Fig. 3-11 demonstrates a network in which all users’ SNR levels are very low such that the SINRs of the selected users are too low to activate the AMC mechanism. If the channel conditions are always poor, users will always be assigned Mode 1, i.e. AMC mechanism is not activated. This makes the proposed method acts almost the same as the Suboptimal AMC (1). To deal with the poor channel condition cases, some more robust modes, e.g. smaller constellation size or lower coding rate, may be considered to be added into the AMC mechanism to combat the bad environments. Besides, there still exist some losses in the low traffic area compared to the Suboptimal AMC (2) curve. This is because that although the SNR levels are low, the actual interferences from other users in the low traffic region are low as well.
Fig. 3-11 Throughput comparison between Joint AMC and suboptimal methods of Case 2 (SNR_max = 13 dB, SNR_min = 9 dB)
Case 3:
This case simulates an environment in which all users’ SNR levels are high. It can be observed that the trends of Fig. 3-12 (Case 3) and Fig. 3-11 (Case 1) are almost the same, except that Case 3 corresponds to a better environment, which leads to a higher attainable channel capacity than others. In this kind of environment, larger constellation size modulation and higher coding rate transmission modes may be added to further increase the attainable channel capacity.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0 0.5 1 1.5 2 2.5 3
Packet generating probability
Effective throughput (Packet/Slot)
Joint AMC
Suboptimal AMC(1) Suboptimal AMC(2)
Fig. 3-12 Throughput comparison between Joint AMC and suboptimal methods for Case 3 (SNR_max = 30 dB, SNR_min = 26 dB)
Case 4:
This case demonstrates a network in which there are 24 users. It can be observed that the trend in Fig. 3-14 acts like a left-shifted version of the other cases. Since the number of users in this case is larger, the probability of the selected accessing users being idle is lower. Therefore the low traffic loss problem is less severe because the average waiting slots of users becomes longer when there are many users in the network. The longer the users wait the more probable they have packets to send as they are selected to access the channel; this mitigates the low traffic loss problem.
Fig. 3-13 Network deployment of 24 users in grid distribution
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fig. 3-14 Throughput comparison between Joint AMC and suboptimal methods for Case 4 (M = 24, SNR_max = 28 dB, SNR_min = 16 dB)
Case 5:
In typical AMC mechanisms, there are usually more than two modes for selection. The reason we only adopt two modes in this thesis is that using the two modes are quite enough for us to clarify the main idea of the proposed method. In this case, we add two more modes into the AMC mechanism as in Table 3-2 to characterize a more realistic system environment. In Fig. 3-15, it can be observed that attainable channel capacity is further increased, and the Joint AMC still outperforms other suboptimal methods. However, the low traffic loss becomes more severe since the coding rate of Mode 4 is higher.
Table 3-2 Adopted AMC modes in Case 5
Mode 1 2 3 4
Modulation BPSK QPSK 16-QAM 64-QAM
Coding rate 1/3 1/2 3/4 7/8
Fig. 3-15 Throughput comparison between Joint AMC and suboptimal methods for Case 5
3.5 Summary
In this chapter, we give a detailed description of the proposed Joint AMC-MAC design. By the two simulations in Section 2.1, we showed that if the AMC mechanism is directly incorporated into MPR environments, the packet reception capabilities of the physical layer could not be fully exploited by the suboptimal flow depicted in Fig.
3-2. The Joint AMC-MAC algorithm uses a simple iterative procedure in Fig. 3-7 that tries every possible combination of user selection and modes to find the optimal user-mode combination, which can fully exploit the physical layer’s MPR capabilities.
However, it can be observed that there exist some losses in the low traffic region in all simulated cases. This low traffic loss problem will be tackled in the next chapter, in which an enhanced version of the Joint AMC-MAC design will be introduced.
Chapter 4
Enhanced Joint AMC-MAC Design
In this chapter, we introduce the proposed Enhanced Joint AMC-MAC (Enhanced JAM) design. First, the proposed Probability Based Mode Tuning (PBMT) method is introduced. The Enhanced Joint AMC-MAC design is an extended version of the Joint AMC-MAC design discussed in Chapter 3. By exploiting the advantage of low complexity in MGPQ and throughput enhancement due to the joint optimization of user selection and modes assignment in the Joint AMC-MAC design, the Enhanced Joint AMC-MAC design further takes the traffic conditions into consideration during the iterative procedure. By the PBMT method, some users who originally assigned as Mode 1 may have chances to be reassigned as Mode 2 because the actual interference levels in the lower traffic region may be low. This method can be used to tackle the low traffic loss problem in Chapter 3. The simulation results show that the Enhanced Joint AMC-MAC design further improves the overall network throughput in the low traffic region. In Section 4.1 and 4.2, the proposed PBMT method and the Enhanced Joint AMC-MAC design are introduced. In Section 4.3, the proposed Enhanced Joint AMC-MAC design with MGPQ MAC protocol is summarized. Some numerical results are shown in Section 4.4. Section 4.5 summarizes this chapter.
4.1 Proposed Probability Based Mode Tuning Method
In the previous chapter, the Joint AMC-MAC algorithm is discussed. However, the simulation results show that there exist some throughput losses in the low traffic region. The low traffic loss results from the improper full load assumption which makes some of the selected users may be assigned robust but low coding rate mode to combat the overestimated interferences. To solve this problem, we can further utilize the packet generating probability and two properties of MGPQ protocol, the waiting slot counter and flag-bit. By these traffic related parameters, we can estimate the probability of the selected user being active, i.e. the interfering probability to other users. With this probability, we can obtain a new SINR level which is closer to the actual SINR level in the low traffic region. After the new SINRs being obtained, some users originally assigned as Mode 1 may have chances to be reassigned as Mode 2, which can compensate the low traffic loss problem in the previous chapter.
The modified SINR calculation (2.9) is modified as
( ( ))
represents the probability of user j having packet to transmit, i.e. the probability of user j to interfere other users. p is the packet generating probability and Wj is the number of waiting slots of user j. And (3.3) becomes
( )
( ) ( )With (4.3), we calculate (3.5) and (3.6) again and update the corresponding mode vector (i.e. the
n
0,AMC( U t
( ))
th row ofCM
AMC( U t
( ))
) to make sure that these users’ modes can match more to the actual traffic conditions..4.2 Proposed Enhanced Joint AMC-MAC Algorithm
In this section, the Enhanced Joint AMC-MAC algorithm is introduced. Fig. 4-1 depicts the Enhanced Joint AMC-MAC algorithm, which is a slightly modified version of the Joint AMC-MAC algorithm in Fig. 3-7. In the iterative procedure, we still use the full load assumption to obtain the optimal user-mode combination that maximizes the average throughput. Then, we append the PBMT algorithm after the iteration part to perform the mode reassignment procedure. The Enhanced Joint AMC-MAC not only compensates the low traffic loss discussed in the previous chapter, but also remains the advantage of the Joint AMC-MAC algorithm in the heavy traffic region since the estimation of users’ activity is more accurate. Some numerical results are shown in the next section.
Fig. 4-1 Enhanced Joint AMC-MAC algorithm
4.3 Proposed Enhanced Joint AMC-MAC Design with MGPQ MAC Protocol
Central controller:
I. Put all users in the user set into the PREM group.
II. Input: U(t)
temp_n0 = 1; temp_ η = 0;
η
AMC( U t
( ))
= 0while temp_n
0 <= Ma) Calculate the SINR levels of the first temp_n0 users in U(t) by (2.9).
b) Assign modes to the first temp_n0 users and record the associated average throughput (temp_ η ) by the corresponding SINRs.
if temp_ η > η
AMC( U t
( ))
(
( ))
AMC
U t
η
= temp_ η(
( ))
0,AMC
n U t = temp_n
0Record the current user-mode combination
end if
temp_n0 = temp_n0 + 1
end while
III. Recalculate the selected
n
0,AMC( U t users’ SINR by (4.1) and update the
( ))
corresponding transmission modes.IV. Select first
n
0,AMC( U t users (by the order of PREM, ACTIVE, and then
( ))
STANDBY group) in the user set and adopt the updated corresponding transmission modes to access the channel.a) If the packet of a certain user is received successfully, then put the user to the tail of the ACTIVE (if the flag bit is on) or STANDBY group (if the flag-bit is off). Reset its count of waiting slots to zero.
b) If, for a certain user, the buffer is empty (no packet sent) or there is packet transmitted but not successfully received, then put the user back to the tail of the STANDBY or ACTIVE group in which the user originally stayed. Reset its count of waiting slots to zero.
V. Increase waiting slots of all users in the user set by one.
VI. Move those users with waiting slots equal to S to the PREM group.
VII. Repeat steps II to VI.
4.4 Computer Simulations
In this section, we simulate the proposed Enhanced Joint AMC-MAC method and compare it with the proposed Joint AMC-MAC method and two suboptimal methods mentioned in Chapter 3. The network and the corresponding parameters are defined the same as in Chapter 3.
Case 1:
In the low traffic region of Fig. 4-2, it can be observed that the Enhanced Joint AMC curve behaves like the Suboptimal AMC (2) curve because the PBMT method reassigns some selected users as Mode 2. The mode reassignment procedure leads to the compensation of the low traffic loss in Fig. 3-9. Besides, in the heavy traffic region, the Enhanced Joint AMC curve overlaps the Joint AMC curve and attains the channel capacity, which means that the MPR capabilities can still be fully exploited by the Enhanced Joint AMC-MAC algorithm.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fig. 4-2 Throughput comparison among Enhanced JAM and other methods for Case 1
Case 2:
As in Fig. 4-2, the Enhanced Joint AMC curve in Fig. 4-3 behaves the same as the Suboptimal AMC (2) curve at low traffic, and then overlaps the original Joint AMC curve with increasing traffic load. Since low traffic loss is not poor in such an environment, the improvement of the Enhanced Joint AMC over the Joint AMC in the low traffic region is not obvious too.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0 0.5 1 1.5 2 2.5 3
Packet generating probability
Effective throughput (Packet/Slot)
Enhanced Joint AMC Joint AMC
Suboptimal AMC(1) Suboptimal AMC(2)
Fig. 4-3 Throughput comparison among Enhanced JAM and other methods for Case 2
Case 3:
Since the users’ SNRs are quite high in this case, the Enhanced JAM curve not only recovers the losses, but also slightly outperforms the Suboptimal AMC (2) curve, which is different from Case 1 in which the Enhanced JAM curve in Fig. 4-2 just acts the same as the Suboptimal AMC (2) curve in the low traffic region.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fig. 4-4 Throughput comparison among Enhanced JAM and other methods for Case 3
In the following Case 4 and Case 5, the trends of the curves are basically the same as in previous figures. The Enhanced JAM curve firstly recovers the losses in the low traffic region, and then overlaps with the original Joint AMC curve in the heavy traffic region.
Case 4:
E ff e c ti v e t h roughput (P ac k e t/ S lo t)
Enhanced Joint AMCJoint AMC
Suboptimal AMC(1) Suboptimal AMC(2)
Fig. 4-5 Throughput comparison among Enhanced JAM and other methods for Case 4
Case 5:
Fig. 4-6 Throughput comparison among Enhanced JAM and other methods for Case 5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fig. 4-7 Loss recovery of Enhanced Joint AMC compared with Suboptimal AMC (3) and (4) for Case 5
4.5 Summary
The proposed Enhanced JAM design is described in this chapter. This design incorporates the proposed PBMT method to recover the low traffic throughput loss of the original Joint AMC-MAC design. By appending the PBMT method at the tail of the JAM design, the Enhanced JAM provides opportunities to the selected users to be reassigned higher order transmission modes, which may be more proper to the actual traffic conditions. Computer simulations show that throughput performance can be improved by assigning users with more suitable modes instead of improper modes based on overestimated interferences.
Chapter 5
Conclusions and Future Works
In the beginning, this thesis introduces the basic idea of the AMC mechanism and reviews the MAC protocol with MPR capability called MGPQ, which is proposed with a simple flag-assisted mechanism and an efficient multi-priority user grouping strategy to achieve high performance in a wireless network with MPR capability.
However, if the AMC mechanism is directly incorporated into the MPR environment, the MPR capability may not be fully utilized during the MAC scheduling process since the conventional MPR matrix cannot reflect the dynamic properties of the physical layer. Our design objective is to perform a cross layer design by jointly optimizing the user selection and the associated transmission modes, which makes the MPR capability fully exploited during the scheduling process. The proposed Joint AMC-MAC algorithm and Enhanced Joint AMC-MAC algorithm achieve the objective and improve the system throughput performance.
Aiming at fully exploiting the MPR capability of the AMC incorporated physical layer, this work tries to perform a cross layer design in a direction from MAC to PHY instead of a traditional way of PHY to MAC. By the proposed JAM design introduced in Chapter 3, we firstly find the optimal user-mode combination that attains the maximum average throughput from MAC layer, then reversely informs PHY of which
transmission mode should be adopted to fully utilize the MPR capability. In the JAM design, we use the full load assumption of traditional MPR works, which means that all the selected users are assumed to be active to transmit packets. This assumption may overestimate the interference levels of the selected users such that some users are assigned robust but low coding rate mode to combat the nonexistent interferences, i.e.
the corresponding modes are underestimated. Therefore, the optimal user-mode combination obtained in the proposed method is inherently optimal for heavy traffic conditions, where the full load assumption is more likely to be valid. In the low traffic region, the improper full load assumption may cause some throughput loss due to users’ underestimated transmission modes.
In Chapter 4, the PBMT method of the Enhanced JAM algorithm is proposed to recover the low traffic throughput loss by providing the selected users with opportunities to be reassigned more proper modes to the associated traffic conditions.
There are three steps in the PBMT. We firstly estimate every selected user’s probability of being active, i.e. the probability of interfering other selected users.
Secondly, we recalculate the selected users’ SINRs by multiplying the interferences with the estimated active probabilities. The third step is to check whether the updated SINRs are qualified to be reassigned higher order modes. It is demonstrated that the proposed method outperforms other suboptimal methods, in which the AMC mechanism is directly incorporated into MPR environments. The simulations in Sections 3.4 and 4.4 show the superiority of the proposed methods.
If AMC mechanism is directly incorporated into MPR environments, then the users’ SINRs for mode assignment depend on the number of selected users, i.e. n0. However, the information about n0 needs to be extracted from the MPR matrices, which are constructed based on users’ parameters of transmission modes. Since the
traditional MPR matrix cannot reflect the channel dynamics in the communication networks, the extracted information may not be correct. This may cause the user selection and the corresponding modes non-optimal and make it impossible to fully exploit the MPR capability of the physical layer. The key reason for the better performance of the proposed methods is that we use another approach to find the user-mode combination that can achieve the maximum average throughput, or channel capacity. We have also used the simulation results to justify the effectiveness of the proposed method. Comparing with suboptimal methods, the throughput enhancement is shown in Section 3.4 and 4.4.
There are still some issues remaining to be further investigated in this work. First, how to efficiently determine the optimal user-mode combination is a concern. Also, the estimation of users’ active probability could be modified to further achieve a higher accuracy. Furthermore, the extension of the proposed MAC protocols to different systems such as OFDMA systems is a subject worthy of investigation.
Bibliography
[1] S. Shakkottai, T. S. Rappaport, "Cross-layer design for wireless networks,"
IEEE Communications Magazine, vol. 41, pp. 74-80, Oct. 2003.
[2] Q. W. Liu, S. Zhou, and G. B. Giannakis, "Cross-layer combining of adaptive modulation and coding with truncated ARQ over wireless links," IEEE
Transactions on Wireless Communications, vol. 3, pp. 1746-1755, Sep. 2004.
[3] V. Srivastava and M. Motani, "Cross-layer design: A survey and the road ahead," IEEE Communications Magazine, vol. 43, pp. 112-119, Dec. 2005.
[4] X. J. Lin,N. B. Shroff, and R. Srikant., "A tutorial on cross-layer optimization in wireless networks," IEEE Journal on Selected Areas in Communications, vol. 24, pp. 1452-1463, Aug. 2006.
[5] Q. W. Liu,X. Wang, G. B. Giannakis., "A cross-layer scheduling algorithm with QoS support in wireless networks," IEEE Transactions on Vehicular
Technology, vol. 55, pp. 839-847, May 2006.
[6] H. Su and X. Zhang, "Cross-layer based opportunistic MAC Protocols for QoS provisionings over cognitive radio wireless networks," IEEE Journal on
Selected Areas in Communications, vol. 26, pp. 118-129, Jan. 2008.
[7] A. K. Karmokar and V. K. Bhargava, "Performance of cross-layer optimal adaptive transmission techniques over diversity Nakagami-m fading
[7] A. K. Karmokar and V. K. Bhargava, "Performance of cross-layer optimal adaptive transmission techniques over diversity Nakagami-m fading