A Novel Packet Scheduling Algorithm For LTE Downlink Transmission Systems

(1)

_______________________________________________________________________________________________

A Novel Packet Scheduling Algorithm For LTE Downlink Transmission Systems

1R Dhaya, ²S Pushpa

1,2Dept. of CSE, St.Peters University, Chennai Abstract. The release 8 of 3GPP project which marked the

beginning of LTE, implemented a lot of novel and efficient changes to the Physical layer of the communication networks. By implementing OFDM for downlink channels, it brought inherent efficiency in communication. But packet scheduling algorithms which are implemented at the eNodeB, actually allocate resources to the User Equipment (UE) based on different parameters used in the PS algorithms. Different parameters like Channel Quality Index (CQI), Quality of Service (QoS), Quality of Experience (QoE), latency, packet loss ratio, throughput were considered when the efficiency of PS algorithms were analyzed. Analyzing the usage of buffers at both the ends of eNodeB and UE will still improvise the performance of PS algorithms. This paper proposed a Modified Resource Allocation Algorithm that assigns priorities to UE while allocating resources based on CQI and buffer statuses to ensure fairness and get optimum throughput out of the system. Finally the proposed Modified Resource Allocation Algorithm is simulated against existing PS algorithms to demonstrate its performance.

Keywords: LTE, Packet Scheduling, QoS, CQI , eNode B

I. INTRODUCTION

With the advent of next generation technologies and networking, the field of wireless networking is moving at a very fast rate to keep up with the changing needs and requirements of the end users [1]. After the 3GPP release, which is the starting point for NGN, the networking world has a great challenge: satisfying the needs of 4.2 billion mobile phone users. Gone are the days when users use their phone just to place a call or send a message [2]. The entire functionality of a desktop computer can be done using a mobile phone and that has added a huge demand on the shoulders of the service providers [4]. The current 3G technology continues to use circuit switching technology for packet sending and reception. But it is undesirable, since one can experience higher HOL delays because of that. So the NGN should have a stable packet scheduling algorithm, with backward compatibility with the existing networks too.

This has led to the evolution of 4G Networks [5]

In this paper, we Proposed Modified Adaptive Resource Allocation which is jointly considering the user scheduling process and Resource Blocks allocation to improve overall system throughput, Energy Efficiency Latency and provide QoS guarantee to keep certain

fairness among users in LTE downlink transmission systems. For user scheduling, each user’s queue priority is ranked according to its remaining life time or its queue overflow probability which is estimated by applying large deviation principle [3]. In the aspect of RBs allocation for adjusting the service rates is proposed based on the user queues’ priorities and dynamically allocates RBs in order to avoid buffer overflow and satisfy QoS requirement.

II. LITERATUREREVIEW

Huda Adibah Mohd et al (2009) investigated the performance of packet scheduling algorithms developed for single carrier wireless systems from a real time video streaming perspective and the performance evaluation was conducted using the downlink third generation partnership project long term evolution system as the simulation platform. Samia Dardouri (2009) evaluated performance of the performance of six scheduling algorithms suggested for LTE downlink transmission and the analysis was inspected for a multicellular with interference scenario for real-time and non-real- time traffic.

These efficient scheduling schemes, aiming to get better overall system performance. The frame level scheduler (FLS) algorithms outperform other, by balancing the QoS requirements for multimedia services. Most of the basic algorithms were compared for a single cell scenario with 20-200 users who have a mobility of 60- 120 km/h..Pardeep Kumaret al(2016) covered the performance analysis of Proportional Fair (PF), Modified Largest Weighted Delay First (M-LWDF) and Exponential-Proportional Fairness (EXP-PF) algorithms, carried out on a single cell affected by interference for different flows such as VoIP, VIDEO and Best Effort.

Pardeep Kumaret al(2016a) discussed the frequency reuse and interference problems of LTE.The algorithms compared were PF, Exp-PF and M-LWDF for video and CBR flows classified the algorithms based on the way RBs are allocated: channel unaware, channel aware/QoS unaware and channel aware/QoS aware. Ruiyi Zhu et al(2016) proposed a buffer-aware adaptive resource allocation scheme for LTE downlink transmission to improve the overall system throughput while providing statistic QoS guarantee and keep certain fairness among

(2)

_______________________________________________________________________________________________

users. Yan Lin(2008), Guangxin Yue(2008) proposed a channel-adapted and buffer-aware scheduling algorithm in LTE wireless communication system. The algorithm was proposed on allocation of RBs based on CQI and prioritizing the traffic as RT or NRT. Kumbesan Sandrasegaran et al(2010) proposed a new packet scheduling algorithm for real time (RT) traffic in downlink third generation partnership project long term evolution (3GPP LTE) system .The proposed algorithm utilized each user's packet delay information and its instantaneous downlink channel conditions when making scheduling decisions. Qurat-ul-Ain et al(2015) proposed an algorithm for those users who move at very high speeds. If a user moves at various speeds, the variance of the CQI is very high. If this is above a specific threshold, then the next location of CQI is predicted or PF algorithm is implemented. Moustafa M.

Nasralla et al(2013) proposed a strategy for resource allocation for different traffic classes at the Medium Access Control (MAC) layer of wireless systems based on Orthogonal Frequency Division Multiple Access (OFDMA), such as the recent Long-Term Evolution (LTE) wirelessstandard.

III. EXISTINGSYSTEM

Orthogonal Frequency-Division Multiple Access (OFDMA) is adopted in LTE downlink transmission systems which allow high flexibility in the resource allocation among the potential users[6].Due to the diverse channel qualities of the users as well as Quality- of-Service (QoS) requirement, designing a multi-user resource allocation scheme in LTE downlink transmission systems to improve the total system throughput, guarantee QoS requirements and achieve fairness still be an interesting and challenging problem [8]. Resource allocation in wireless data systems has been discussed in several related works. A resource allocation scheme in LTE systems is presented for reducing the waste of system resource and improving the total throughput by utilizing the knowledge of buffer status and channel conditions [7]. However, these algorithms have not considered the system fairness which plays an important role in the system performance [9]. In fact, providing fairness among users is an essential design consideration although it usually sacrifice the system throughput and/or violates QoS requirements. Proportionality Fair algorithm achieves a tradeoff between system throughput and fairness [10]. PF metric algorithm, which introduces the status of queues into PF metric.

However, it is pointed out that PF metric may result in an inefficient Resource Block (RB) assignment, because a single RB is considered in isolation one-by-one regardless of other RB assignment status. By considering both the constraint of finite buffer [11] space and fairness, Channel-Adapted and Buffer-Aware (CABA) packet scheduling algorithm which applies the user priority in the resource allocation to avoid buffer overflow. However, it is very different to choose the empirical parameters in the priority function appropriately. It may induce excessive resource

allocated to the users, which reduces the systemutility.

IV. PROPOSEDSYSTEM

The Proposed Modified Adaptive Resource Allocation Algorithm is an adaptive resource allocation scheme by jointly considering the user scheduling and Resource Blocks allocation to improve overall system throughput, Energy Efficiency Latency and provide QoS guarantee to keep certain fairness among users in LTE downlink transmission systems. For user scheduling, each user’s queue priority is ranked according to its remaining life time or its queue overflow probability which is estimated by applying large deviation principle. In the aspect of RBs allocation, an online measurement based algorithm for adjusting the service rates is proposed based on the user queues’ priorities and dynamically allocates RBs in order to avoid buffer overflow and satisfy QoSrequirement.

V. SYSTEMDESIGN

System architecture is a high level diagram depicting the relationship between the system entities in the system.

Figure 1 depicts the high level system architecture of the proposed system design. A single cell can contain multiple UEs of which some maybe mobile and some may be stationary. These cells report the CQI, SINR values at every TTL based on which the resources to be allocated are understood by the packet scheduler at the eNB. Every UE has a buffer, typically a queue, based on which the allocation is done. Every UE is allocated a priority depending on the values of CQI, SINR, buffer length and priority of the data. This decision of priority allocation and RB allocation is done as per the algorithm at the eNB packet scheduling.

Figure 1: High level architecture of the proposed system Design

VI. MODIFIED RESOURCEALLOCATION ALGORITHM

Suppose there are n user queues indexed by the set Φ = {1, 2,...K} and N SBs indexed by the set Ω = {1, 2,...N}. The detail of the strategy is presented in Algorithm

while Φ ƒ= ∅ or Ω ƒ= ∅ do

(3)

_______________________________________________________________________________________________

if the value of Rn(t) exists then

Choose the user n1 = arg mink∈Φ {Rn(t)} else

Choose the user n1 = argmaxkΦ{Pt+T } endif

Seek the SB n1 = arg maxn∈Ω{γn,n1 } for the k1th user while rn1 (t) ≤ An (t)do n1 1

Ω = Ω\{n1}

Seek the SB n1 = arg maxn∈Ω{γn,k1 } end while after the data of user k1 has been transmitted, set Φ = Φ\{n1}

end while

VII. PERFORMANCEEVALUATION

Modularity of a paper determines the success of the system. Modularity helps in building a cohesive and coupled system which will work as expected, in an effective way. The modules in the proposed system are:

1. End User Environment

2. Experiment and Performance Metrics

3. Comparing the various Performance metrics of Diﬀerent Algorithms(PF, Ex-PF,M-LWDF)

4. Proposed Algorithm- Modified Adaptive Resource Allocation Algorithm

5. To improve the total system while guaranteeing the certain fairness among users and reducing the average Bit Lose Rate(BLR)

End User Environment: This module helps to design and finalize the end user environment for the proposed system. This module finds the Number of UEs to be deployed, Number of eNBs, Number of cells, The type of traffic (GBR, non-GBR), The type of application Based on these the nodes are created and deployed.

Experiment and Performance Metrics: This module suggests toanalyze the performance metrics

- The performance parameters that have been considered

- The way the UEs areprioritized - The allocation ofRBs

o Based on these data, demerits in the algorithms can be identified, and they can be tried to rectify in the proposed system.

Comparing the various Performance metrics of Diﬀerent Algorithms (PF, Ex-PF, M-LWDF):

This module is to compare the performance of the various algorithms based on various parameters. The performance metrics are Packet Delivery Ratio, Throughput, Energy Consumption and Latency

Proposed Algorithm- Modified Adaptive Resource Allocation Algorithm: This module helps in designing the algorithm and ensures that the end user’s QoS and QoE are improved. This algorithm should also ensure that the key parameters are considered and the performance is considerably improvised. This algorithm prioritizes the User Environments (UE) based on their buffer status and the instantaneous data rates at each Time to Live (TTL). At each TTL, UEs that have less power or the UEs that have Guaranteed Bit Rate (GBR) flows are given more priority. But each UE is given a minimal Resource Blocks (RB) count, so as to ensure best effort flows, and minimize starvation

VIII. EXPERIMENTSETUPAND PERFORMANCEMETRICS

We simulated a multiuser scenario, where the maximum number of communicating users was set to n = 10. Here, the bit arrival rate for each user is assumed to obey the Poisson distribution withλ>0.CQIisdiscretizedinto15 levels,which results in 15 diﬀerent pairs of modulation choice and code rate. This implies that there may be 15 possible transmission rates. A mapping between SINR ranges and CQIs is presented in [10]. The obtained CQIs are then used, together with the number of allocated RBs, to determine transmission rates that are used by the proposed optimization algorithm to improve the overall throughput. To evaluate the performance of the proposed dynamic resource allocation, we deﬁne three metrics as follows:

PDR: This metric is measured using delivery Ratio, which is widely applied for evaluating the system performance. It is described as F (t) = k=1 Dk(t)) k=1 D2 , where F (t) denotes the delivery at time t. Then, the system k(t)delivery ratio can be calculated as F = 1 . T0+Δ F (t). Δ+1t=T0

Figure 2 shows the Average PDR of existing and Proposed System.

Figure 2: Average Packet Delivery Ratio Average throughput: The larger average system throughput impliesbetter performance. Figure3shows the comparison result of Number of Nodes vs. Throughput.

It measures the amount of data that has been

(4)

_______________________________________________________________________________________________

successfully passed through the network. Again, Proposed MARA showed highest throughput, but when the number of users increased, there was a drop in the value, even though it was still higher than theothers.

Figure 3: Average Throughput

Average Energy: The amount of energy needed for allocation of the RBs is also equally important. Figure 4 shows the comparison result of Number of Nodes vs.

Energy Consumption. Of the algorithms, Proposed MARA showed high energy consumption while allocation of resources to the UEs.

Figure 4: Average Energy Consumption Average Latency: Another important metric is the time taken for allocation of the resources (which means time taken for calculation of the metric and finding the priorities for the UEs).Figure 5 shows the comparison results of Number of Nodes vs Latency. It can be seen that MARA has a higher latency than the others and PF has least latency.

Figure 5:Avergy Latency Performance Comparison for Different User Index

We used Network Simulator 2 for implementing the simulations. The corresponding simulation parameters are listed in Table I.

Table 1. Simulation parameters.

In Fig. 6, we show that the average BLR and the average throughput corresponding to 10 users for the ﬁve resource allocation schemes. The X axis denotes the user index. As we can see, in Fig. 6 , the proposed algorithm achieves better performance with average BLR of the existing methods

Figure 6. :Average BLR

Figure 7, the average throughput for each user in the proposed algorithm signiﬁcantly The reason for these is that, we calculate the priority of each user queue by using the remaining life time or queue overﬂow probability, and then allocate RBs dynamically.

Consequently, it achieves a lower value of the average BLR, improves all users’ service rate and keeps a high fairness among all users.

Figure 7: Average Throughput

IX. CONCLUSION

(5)

_______________________________________________________________________________________________

This paper presents the downlink packet scheduling algorithm which is called Modified Resource Allocation Algorithm and its properties.

The work compared the PF, Exp-PF and M-LWDF algorithms for performance metrics like throughput, PDR, latency, energy consumption. The traffic was CBR, and the UEs were assumed to be stationary. It was found that the MARA algorithm has highest performance when compared to the rest. The proposed system deﬁnes the life time of a queue, and its estimation model. According to the users’ queue priority, Modified Adaptive Resource Allocation has been projected to schedule RBs dynamically for adjusting the service rate of the user queues. The proposed algorithm has a better tradeoﬀ among throughput, QoS and fairness. It improves the total system throughput while guaranteeing certain fairness among users and reducing the average BLR.

REFERENCES

[1] A. Bin Sediq, R. H. Gohary, and H.

Yanikomeroglu: Optimal tradeoﬀ between eﬃciency and jain’s fairness index in resource allocation. In Proc. IEEE PIMRC, Sydney (2012)577-583.

[2] A. Toskala, H. Holma, K. Pajukoski, and E.

Tiirola: UTRAN Long Term Evolution in 3GPP.

In Proc. IEEE PIMRC, Helsinki (2006) 1-5.

[3] Andrews, Matthew, and Lisa Zhang: Scheduling algorithms for multicarrier wireless data systems.

IEEE Trans. Networking , vol.19. no. 2.

(2011)447-455.

[4] B.P.S.Sahoo, Deepak Puthal, Satyabrata Swain, Sambit Mishra, “A comparative analysis of packet scheduling schemes for multimedia services in LTE networks”, International Conference on Computational Intelligence and Networks, pp. 110 - 115,2015.

[5] C. Wang and X. Li: A buﬀer-aware resource allocation scheme for 4G LTE systems. In Proc.

IEEE ISCE, Hsinchu (2013)157-158.

[6] E.U.T.R.Access: FurtheradvancementsforE- UTRA physical layer aspects. 3GPP TR 36.814, Tech. Rep. (2010).

[7] E. U. T. R. Access: Physical layer procedures (release 10).mphTS,vol.36.(2011).

[8] F. Capozzi, G. Piro, L. A. Grieco, G. Boggia, and P. Camarda: Downlink packet scheduling in LTE cellular networks: Key design issues and a survey. IEEE Commun. Surveys and Tutorials, vol. 15. no. 2. (2013) 678–700.

[9] G. Y. Yan Lin: Channel-adapted and buﬀer-aware packet scheduling in LTE wireless communication system. In Proc. IEEE WICOM, Barcelona (2008)239-243.

[10] Huda Adibah Mohd Ramli, Riyaj Basukala, Kumbesan Sandrasegaran, Rachod Patachaianand, “Performance Of Well Known Packet Scheduling Algorithms In Downlink 3GPP LTE system”, Proceedings of the 2009 IEEE 9th Malaysia International Conference on Communications,pp.815-820,2009.

[11] I.-F. Chao and C.-S. Chiou: An enhanced proportional fair scheduling algorithm to maximize QoS traﬃc in downlink OFDMA systems. In Proc. IEEE WCNC, Shanghai (2013) 239-243.

[12] Kumbesan Sandrasegaran, Huda Adibah Mohd Ramli and Riyaj Basukala , “Delay-Prioritized Scheduling (DPS) for Real Time Traffic in 3GPP LTE System”, IEEE Wireless Communications and Networking Conference (WCNC), pp. 1- 10,2010.

[13] M. Mandjes: Large deviations for Gaussian queues: modelling communication networks.

John Wiley & Sons, (2007).

[14] Moustafa M. Nasralla and Maria G. Martini, “A Downlink Scheduling Approach for Balancing QoS in LTE Wireless Networks”, IEEE 24th International Symposium on Personal Indoor and

Mobile Radio Communications

(PIMRC),pp.1571-1575,2013.

[15] Pardeep Kumar, Sanjeev Kumar, “Performance Analysis Of Downlink Packet Scheduling Algorithms in LTE Networks”, Advanced Computing and Communication Technologies, Advances in Intelligent Systems and Computing,pp.453-459,2016.

[16] Pardeep Kumar, Sanjeev Kumar, Chetna Dabas ,

“Comparative Analysis Of Downlink Schedluing Algorithms for a cell affected by interference in LTE Network”, Springer, volume 3, Issue 2, pp 135– 153,2016.

[17] Qurat-ul-Ain, Syed Riaz ul Hassnain, Mudassir Shah and Sahibzada Ali Mahmud, “An Evaluation of Scheduling Algorithms in LTE based 4G Networks”, IEEE International Conference on Emerging Technologies (ICET),pp.111-119,2015.

[18] Ruiyi Zhu, Jian Yang, and Peng Si ,” Adaptive Resource Allocation in LTE Downlink Transmission Systems”, Lecture Notes in Electrical Engineering, Springer, Volume 391, pp3-12,2016.

[19] Samia Dardouri, Ridha Bouallegue ,”

Comparative Study For Scheduling Algorithms For LTE Networks”, International Journal of Computer, Electrical, Automation, Control and Information Engineering, Volume 82, Issue 3, pp1405–1418,2014.

(6)

_______________________________________________________________________________________________

[20] Yan Lin, Guangxin Yue, ”Channel-Adapted and Buffer- Aware Packet Scheduling in LTE Wireless Communication System”, IEEE 4th

International Conference on Wireless Communications, Networking and Mobile Computing, pp.12-16,2008.

