Performance Evaluation of Scheduling Strategies for LTE Networks in Downlink Direction Stojan Kitanov
Toni Janevski
University for Information Science and Technology “St. Paul the Apostle” - UIST Ohrid, Republic of Macedonia
[email protected]
Faculty of Electrical Engineering and Information Technologies - FEIT Ss Cyril and Methodius University - UKIM Skopje, Republic of Macedonia
[email protected]
Abstract—LTE (Long Term Evolution) networks have been introduced in 3GPP (Third Generation Partnership Project) release 8 specifications, with an objective of high-data-rate, lowlatency and packet-optimized radio access technology. The QoS for multimedia services are defined in LTE specification. However the scheduling algorithms for supporting real time and non-real time application services are not specified. Therefore different scheduling strategies are proposed. This paper explores the performances of LTE Network for different proposed scheduling algorithms such as Proportional Fair (PF), MaximumLargest Weighted Delay First (M-LWDF), and Exponential Proportional Fairness (EXP/PF) by using the LTE-Sim network simulator. The evaluation is considered for a single cell with interference scenario for different flows such as video, best effort, and VoIP. The performance evaluation is conducted in terms of system throughput, fairness index, delay, and Packet Loss Ratio (PLR). Finally it will be concluded that all 3 scheduling algorithms are suitable for best effort flows, while M-LWDF and EXP/PF are suitable for video flows, and PF and EXP/PF are suitable for VoIP flows. Keywords—Long Term Evolution (LTE); LTE-Advanced; System Architecture Evolution (SAE); Evolved Packet System (EPS); Throughput; Scheduling algorithms; Performance evaluation.
I.
INTRODUCTION
Recently, there has been an increasing demand for Internet access over the mobile devices. To address this, the wireless telecommunication industry defined a new air interface for mobile communications that provides a framework for high mobility broadband services and increase in the overall system capacity. One emerging technology that supports this air interface is LTE (Long Term Evolution), which is the current evolutionary step in the third Generation Partnership Project (3GPP) roadmap for future wireless cellular systems [1]. LTE was introduced in 3GPP Release 8, which—besides the definition of the novel physical layer—also contains many other remarkable innovations [2, 3]. Most notable are redevelopment of the system architecture, now called System Architecture Evolution (SAE); self-organization, and self-optimization of the network; and introduction of home base-stations.
The overall objective for LTE is to provide a significantly increased peak data rates, reduced delay (latency), scalable bandwidth capacity and more multi-user flexibility than the currently deployed networks. It also provides backwards compatibility with existing 3GPP and 3GPP2 technologies (GSM, UMTS, HSPA, CDMA 2000), as well as inter-working with other non-3GPP and non-3GPP2 technologies such as IEEE 802.16 WiMAX (Worldwide Interoperability for Microwave Access), and IEEE 802.11 WLAN (Wireless Local Area Networks), i.e. WiFi (Wireless Fidelity) standard [4]. The LTE system is not only designed for substantial performance enhancements but also to reduce the cost per bit. Taking the technological point of view what LTE concept makes the number one are the increased data rate, improved geographical coverage and reduced latency. On the other side, the commercial point of view of the LTE promises more competitive business case for the mobile operators and the base network for service providers. Since LTE is relatively new standard it is important to model and analyze the behavior and the performances of LTE Network. The Quality of Service (QoS) for multimedia services is defined in LTE specification. However the scheduling algorithms for supporting real time and non-real time application services are not specified. Therefore different scheduling strategies are proposed. In these scheduling algorithms, a priority value is assigned to each connection regardless its type depending on certain criterion. The connection with the best criteria is scheduled first at the next transmission time interval (TTI). This approach has the advantage of low implementation complexity. However it is difficult to define a single priority criterion, due to the different traffic characteristics and QoS requirements of Real Time (RT) and Non Real Time (NRT). Therefore, it is necessary to use different algorithms for each type of service (RT and NRT). This paper explores the performances of LTE Network for different proposed scheduling algorithms such as Proportional Fair (PF) [4, 5], Maximum-Largest Weighted Delay First (MLWDF) [6-9], and Exponential Proportional Fairness (EXP/PF) [9, 10] by using the LTE-Sim network simulator [11, 12]. LTESim is chosen because it is open source network simulator and
Proceedings of the XI International Conference ETAI 2013, 26th -28th of September 2013, Ohrid, Republic of Macedonia
supports these scheduling algorithms. The evaluation is considered for a single cell with interference scenario for different flows such as video, best effort, and VoIP. The performance evaluation is conducted in terms of system throughput, fairness index, delay, and Packet Loss Ratio (PLR). The structure of the paper is organized as follows. Section II presents an overview on LTE Network Architecture, and LTE-Sim network simulator. Section III explains the possible scheduling algorithms in LTE. Section IV defines the performance evaluation scenario for different flows for different scheduling strategies in LTE. Section V presents the simulation results. Finally Section VI concludes the paper and provides information about future work. II.
OVERVIEW OF LTE NETWORK AND LTE-SIM
LTE is a cellular system that supports only PacketSwitched (PS) services, compared to the previous CircuitSwitched (CS) cellular systems. It provides seamless Internet Protocol (IP) connectivity between User Equipment (UE) and the Packet Data Network (PDN), without any disruption to the end users’ applications during mobility. The term ‘LTE’ encompasses the evolution of the radio access through the Evolved-Universal Terrestrial Radio Access Network (EUTRAN). This evolution is accompanied by an evolution of the non-radio aspects of the network under the term ‘System Architecture Evolution’ (SAE) which includes the Evolved Packet Core (EPC) network. Together LTE and SAE comprise the Evolved Packet System (EPS). EPS uses the concept of IP packet flows with a defined Quality of Service (QoS) to route IP traffic from a gateway in the PDN to the UE [4]. An overview of LTE-SAE System (EPS System) architecture is given on the Fig. 1 below, and description of the LTE architecture components can be found in [4]. As it can be seen, the access network of LTE-SAE System, E-UTRAN consists only of eNodeBs. One of the main eNodeB functions in an LTE system is to manage the resource scheduling for both uplink and downlink channels. At the eNodeB the packet scheduler performs a user selection priority, based on criteria such as channel conditions, Head Of Line (HOL) packet delays, buffer status, service type, etc. The eNodeB has complete information about the channel quality by the use of Channel State Information (CSI).
The ultimate purpose of this function is to satisfy the Quality of Service (QoS) requirements of all users by trying to reach, at the same time, an optimal trade-off between utilization and fairness [4, 13]. The QoS aspects of the LTE downlink are influenced by a large number of factors such as: Channel conditions, resource allocation policies, available resources, delay sensitive/insensitive traffic. This goal is very challenging, especially for real time multimedia applications that are characterized by strict constraints on packet delay and jitter. This paper examines the performance evaluation of LTE network by using different scheduling algorithms, which are briefly explained in the next section. LTE-Sim, on the other hand, is a network simulator that encompasses several aspects of LTE networks [11, 12]. It includes both the Evolved Universal Terrestrial Radio Access (E-UTRAN) and the Evolved Packet System (EPS). It supports single and heterogeneous multi-cell environments, QoS management, multi-users environment, user mobility, handover procedures, and frequency reuse techniques. The following network nodes are supported: User Equipment (UE), evolved Node B (eNB), Home eNB (HeNB), Home eNB (HeNB), and Mobility Management Entity/Gateway (MME/GW). Different types of traffic generators at the application layer have been implemented and the management of data radio bearer is supported. Finally, well-known scheduling strategies (such as Proportional Fair, Modified Largest Weighted Delay First, and Exponential Proportional Fair, Log and Exp rules), Adaptive Modulation Coding (AMC) scheme, Channel Quality Indicator feedback, and models for physical layer are supported. Both single cell and multi cell simulation scenarios are supported [11, 12]. III.
POSSIBLE SCHEDULING ALGORITHMS IN LTE
Some of the possible scheduling strategies in LTE Networks are Proportional Fair (PF) [4, 5], Modified-Largest Weighted Delay First (M-LWDF) [6 – 9], and Exponential Proportional Fairness (EXP/PF) [9, 10]. Proportional Fair (PF) scheduling algorithm was initially implemented in High Data Rate (HDR) networks [4, 5]. It compromises between a fair data rate for each user and the total data rate. This algorithm maximizes the total network throughput and guarantee fairness among flows such as video, best effort and VoIP. PF scheduling algorithm is very suitable scheduling for non-real time traffic. It assigns radio resources taking into account both the experienced channel quality and the past user throughput. PF algorithm schedules resources to a user j when its instantaneous channel quality is high relative to its own average channel condition over time: to its own average channel condition over time:
j=
µi (t ) µi
Fig. 1. The LTE-SAE System (EPS System) Architecture [4].
Proceedings of the XI International Conference ETAI 2013, 26th -28th of September 2013, Ohrid, Republic of Macedonia
(1)
where
µi (t ) denotes the data rate corresponding to the channel
state of the user i at time slot supported by the channel.
t , µi is the mean data rate
PF algorithm meets scheduling requirements of non-realtime services. However it is not ideal for delay-sensitive realtime services since it does not consider the delay of data packet. To solve this problem, some researchers proposed Modified Largest Weighted Delay First (M-LWDF) algorithm [6 – 9] and further, others suggested EXP/PF algorithm [9, 10]. Modified-Largest Weighted Delay First (M-LWDF) is a scheduling algorithm that supports multiple real time data users in CDMA-HDR systems [6 – 9] with different QoS requirements. This takes into account instantaneous channel variations and delays in the case of video service. The MLWDF scheduling rule tries to balance the weighted delays of packets and to utilize the knowledge about the channel state efficiently. At time slot t , it chooses user j for transmission as follows:
j = max ai i
where
µi (t ) Wi (t ) µi
(2)
µi (t ) is the data rate corresponding to the channel state
of the user i at time slot t ,
µi
is the mean data rate supported
Wi (t ) is the HOL packet delay and ai > 0 are the weights, which define the required level of QoS for i = 1, 2, …, N. In practice, one rule for choosing the weights ai is:
by the channel,
ai = − log(δ i )Ti
(3)
where Ti is the largest delay that user i can tolerate and δ i is the largest probability with which the delay requirement can be violated [14]. Exponential Proportional Fairness (EXP/PF) scheduling algorithm [9, 10] supports multimedia applications in an Adaptive Coding Modulation and Time Division Multiplexing (ACM/TDM) system, which means that a user can belong to a real time service or non-real time service. This algorithm has been designed to increase the priority of real time flows with respect to non-real time ones. At time slot t , the EXP rule chooses user j for transmission as follows:
j = max ai i
µi (t ) a W (t ) − aW exp( i i ) µi 1 + aW
(4)
where all the corresponding parameters are the same as in the
aW defined as
M-LWDF rule, except the term
aW =
1 N
∑ a W (t ) i
i
When the HOL packet delays for all the users do not differ a lot, the exponential term is close to 1 and the EXP rule performs as the Proportionally Fair rule. If for one of the users the HOL delay becomes very large, the exponential term overrides the channel state-related term, and the user gets a priority. In data communication networks and in multiplexing, a division of the bandwidth resources is said to be max-min fair when: firstly, the minimum data rate that a data flow achieves is maximized; secondly, the second lowest data rate that a dataflow achieves is maximized. IV.
SIMULATION SCENARIO AND PARAMETERS
The performance evaluation of PF, M-LWDF, and EXP/PF scheduling algorithms in LTE are performed on a single cell with interference scenario for different flows such as video flows, VoIP flows, and FTP flows. The distribution of the flows among the subscribers is as follows: 40% of subscribers use video flows, 40% use VoIP flows and the remaining 20% use best effort, i.e. FTP flows. Every subscriber moves with a constant speed km/h in random directions (random walk) [15]. LTE-Sim simulator is used to perform this evaluation [11, 12]. According to [4, 13], in the time domain, radio resources are distributed every Transmission Time Interval (TTI), each one lasting 1 ms. Furthermore each TTI is composed by two time slot of 0.5 ms, corresponding to 14 OFDM symbols in the default configuration with short cyclic prefix; 10 consecutive TTIs form the LTE Frame. The simulations are performed on a cell with low density of users, between 5 and 20 users with an increment of 5 users. The channel bandwidth is 10 MHz, and the frame structure is FDD. A video service with 242 kbps source video data rate is used in the simulation available at [16]. For VoIP flows G.729 voice flow are generated by the VoIP application with VoIP bit rate of 8.4 kbps. Best effort flows are created by an infinite buffer application for an ideal greedy source that transmits packets. A summary of the simulation parameters is given in Table I. TABLE I.
LTE DOWNLINK SIMULATION PARAMETERS
Parameters Simulation duration Flows duration Frame structure Radius Bandwidth Slot duration Scheduling time (Transmission Time Interval – TTI duration) Number of Resource Blocks (RBs) Max delay Video bit-rate VoIP bit-rate Minimum number of users Maximum number of users Interval between users
(5)
i
Proceedings of the XI International Conference ETAI 2013, 26th -28th of September 2013, Ohrid, Republic of Macedonia
Value 150 s 120 s FDD 1 km 10 MHz 0.5 s 1 ms 50 0.1 s 242 kbps 8.4 kbps 5 20 5
V.
ANALYSIS OF SIMULATION RESULT
This section analyzes the simulation results. The graphs are divided into three groups: one for the Infinite Buffer (Best Effort flows), one for the Video flows and one for the VoIP flows. All three of them are concentrated on the fairness index, throughput, delay, packet loss ratio [4, 13, 17]. Fairness index (Jain’s Fairness Index) determines whether users or applications are receiving a fair share of system resources [4, 13]. Throughput is the average rate of successful message delivery over communication channel [17]. Delay specifies how long it takes for a bit of data to travel across the network from one node or endpoint to another [17]. Packet loss ratio is the ratio of the lost packets over the number of lost packets and received packets [17].
Fig. 3. Fairness Index for Best Effort Flows.
A. Best Effort Flows The delay for best effort flow is presented on Fig. 2. It can be noticed that the Delay is 1 ms for EXP/PF and very negligible for PF and M-LWDF. In general the delay can be considered to be negligible for all scheduling algorithms. The fairness index for Best Effort Flow is shown on Fig. 3. It can be noticed that Fairness Index is growing up equally for all three algorithms till the point of 15 users. After that the Fairness Index is going down to all users. It means that till the number of 15 users the fair share of the network resources is increasing, and always gets a resource allocation at each TTI. By increasing the number of users above 15 the fair share between the users starts to decrease.
Fig. 4. Packet Loss Ratio for Best Effort Flows.
The packet loss ratio is given on Fig. 4. The packet loss is decreasing constantly as much as the number of users is increasing, which is one of the biggest achievements of the LTE networks. Despite the increase of the users, the loosing of the packets through the transmission is decreased. The throughput performance for best effort flow is given on Fig. 5. The throughput decreases as long as the number of users is increasing, which means that the amount of data that can flow through the network is reduced with the increase of the users. This is very common characteristics for all networks. From all the results given above can be concluded that all 3 algorithms are suitable for best effort flows, since they have nearly the same performances for best effort flows. Fig. 5. Throughput for Best Effort Flows.
B. Video Flows The delay for video flows is given on Fig. 6. It can be concluded that video delay is almost not increasing despite the increase of users for M-LWDF and EXP/PF Scheduling Algorithm. That conclusion does not go for the proportional fair. The delay from the source to the destination of the transmitted videos in the networks (ex. Video Call) is increasing drastically for the proportional fair scheduling algorithm.
Fig. 2. Delay for Best Effort Flows.
The fairness index for video flows is given on Fig. 7. It can noticed that in LTE the fairness index is decreasing as much as the users are getting higher, especially for the proportional fair scheduling algorithm.
Proceedings of the XI International Conference ETAI 2013, 26th -28th of September 2013, Ohrid, Republic of Macedonia
scheduling algorithm. Actually, loosing of the packets through the transmitting process is increased during the video flows. The throughput for the video flows is given on Fig. 9. As the number of users is increases, the throughput also increases for M-LWDF and EXP/PF scheduling algorithms. Therefore, the throughput characteristic of the LTE network when the video signal is transmitted goes up. However this is not the case for the for the proportional fair scheduling algorithms.
Fig. 6. Delay for Video Flows.
Fig. 7. Fairness Index for Video Flows.
Finally it can be concluded that PF scheduling algorithm is not suitable for video flows due to its high delay, packet loss ratio and low throughput. The M-LWDF and EXP/PF scheduling algorithms have better performances for video flows in LTE Networks. C. VoIP Flows The delay for VoIP flows is given on Fig. 10. It can be concluded that the VoIP delay for the LTE users has low and almost unchanged values for PF and EXP/PF scheduling algorithms as the number of users increases. However, as the number of users increases, the delay also increases for the MLWDF scheduling algorithm. This means that delay between the source and destination device for the VoIP flows is used is very stable for PF and EXP/PF scheduling algorithms. The fairness index for the VoIP flows is presented on Fig. 11. It can be noticed that for scheduling algorithms the fairness index for is decreasing till the number of 15 users, and after that it starts with to increase, which means that the fair share of the resources between the devices on the network for the VoIP flows decreases and then after certain number of users it increases again.
Fig. 8. Packet Loss Ratio for Video Flows.
Fig. 10. Delay for VoIP Flows.
Fig. 9. Throughput for Video Flows.
The packet loss ratio for video flows is given on Fig. 8. The packet loss ratio at video transmission is higher than at the buffer transmission, especially for the proportional fair Fig. 11. Fairness for VoIP Flows.
Proceedings of the XI International Conference ETAI 2013, 26th -28th of September 2013, Ohrid, Republic of Macedonia
high delay, packet loss ratio and low throughput. The MLWDF and EXP/PF scheduling algorithms have better performances for video flows in LTE Networks. The most suitable scheduling algorithms for VoIP flows are PF and EXP/PF, while M-LWDF is not suitable due to its very high delay and high packet loss ratio.
Fig. 12. Packet Loss Ratio for VoIP Flows.
LTE is expected to satisfy the market needs for next decade. However, user expectations, traffic growth and new services will demand more and more from the network in the future. ITU has already set a project group to study the future requirements within the framework of the IMT Advanced project. LTE-Advanced already fulfills these requirements. Therefore in future we plan to investigate the performances of the scheduling algorithms for LTE-Advanced, as well as other complex algorithms will be explored. REFERENCES [1] [2] [3] [4]
Fig. 13. Throughput for Video Flows.
[5]
The packet loss ratio for the VoIP flows is shown on Fig. 12. It can be noticed that PF algorithm has the lowest packet loss ratio and M-LWDF has the highest packet loss ratio, independently from the number of the users. However the packet loss ratio also decreases for EXP/PF scheduling algorithm as the number of the users increases. This means that the number of lost packets through the network while transmitting the VoIP signal goes down with the increase of the users. The throughput for the VoIP flows is given on Fig. 13. It can be noticed that throughput for the VoIP flows in LTE for all scheduling algorithms is increasing with the increase of the number of VoIP users.
[6]
[7]
[8]
[9]
[10]
From all the above it can be concluded that it can be concluded that PF and EXP/PF are the most suitable scheduling algorithms for VoIP flows. M-LWDF is not suitable due to its very high delay and high packet loss ratio.
[11] [12]
VI.
CONCLUSION AND FUTURE WORK
This paper evaluated the performance of resource allocation in LTE networks in downlink direction for real time and nonreal time services, i.e. best effort flows, video flows and VoIP flows. Different scheduling algorithms such as Proportional Fair (PF), Maximum-Largest Weighted Delay First (MLWDF), and Exponential Proportional Fairness (EXP/PF) were evaluated by using LTE-Sim network simulator were explored. Finally it was concluded that all 3 algorithms are suitable for best effort flows, since they exhibit nearly the same performances for best effort flows. On the other hand PF scheduling algorithm is not suitable for video flows due to its
[13] [14]
[15]
[16] [17]
(2013) 3GPP, the Mobile Broadband Standard. [Online]. Available: http://www.3gpp.org. (2013) 3GPP Release 8 Standard. [Online]. Available: htpp://www.3gpp.org/Release-8. (2013) 3GPP LTE. [Online]. Available: http://www.3gpp.org/LTE. S. Sesia, I. Toufik, and M. Baker, LTE – The UMTS Long Term Evolution From Theory to Practice 2nd Edition including Release 10 for LTE Advanced, 2nd ed., John Wiley & Sons Ltd., 2011. “Fine particles, thin films and exchange anisotropy,” in Magnetism, vol. III, G.T. Rado and H. Suhl, Eds. New York: Academic, 1963, pp. 271-350. J.-G. Choi and S. Bahk, "Cell-throughput analysis of the proportional fair scheduler in the single-cell environment". IEEE Trans. Veh. Technol., vol. 56, no. 2, pp. 766–778, Mar. 2007. P. Ameigeiras, J. Wigard, and P. Mogensen. "Performance of the m-lwdf scheduling algorithm for streaming services in hsdpa". IEEE Trans. Veh. Technol. Conf., vol. 2, pp. 999-1003, Sep. 2004. Los Angeles, USA. K. Kanghee, K. Insoo, S. Seokjin, etc. Multiple QoS support using MLWDF in OFDMA adaptive resource allocation [J]. The 13th IEEE Workshop on Local and Metropolitan Area Networks, 2004, (4): 217222. A. Matthew, K. Krishnan, R. Kavita. Providing quality of service over a shared wireless link [J]. IEEE Communications Magazine, 2001, 39(2): 150-154. R. Basukala, H. Mohd Ramli and K. Sandrasegaran. "Performance Analysis of EXP/PF and M-LWDF in Downlink 3GPP LTE System". IEEE F. Asian Himalayas Conf., pp. 1-5, Nov. 2009. Kathmandu, Nepal. J. Rhee, J . M. Holtzman, K. Dongku. Performance analysis of the adaptive EXP/PF channel scheduler in an AMsuch C/TDM system [J]. IEEE Communications Letters, 2004, 8(8): 497-499. (2013) Telematics Lab LTE Simulator. [Online]. Available: http://telematics.poliba.it/index.php/en/lte-sim. G. Piro, L. A. Grieco, G. Boggia, F. Capozzi, and P. Camarda, “Simulating LTE Cellular Systems: an Open Source Framework", IEEE Trans. Veh. Technol., vol. 60, no. 2, Feb. 2011, doi: 10.1109/TVT.2010.2091660. E. Dahlman, S. Parkvall, J. Skold, and P. Beming, 3G Evolution HSPA and LTE for Mobile Broadband. Academic Press, 2008. M. Andrews, K. Kumaran, K. Ramanan, A. Stolyar, R. Vijayakumar, and P. Whiting. "Providing quality of service over a shared wireless link". IEEE Communications Mag., vol. 39, no. 2, pp. 150-154, Feb. 2001. B. Jabbari, G. Mason, Y. Zhou, F. Hillier, “Random walk modeling of mobility in wireless networks,” Vehicular Technology Conference VTC 98. 48th IEEE (Volume:1 ), pp. 639 – 643, 1998. (2013) Video trace library. [Online]. Available: http://trace.eas.asu.edu/. D. E. Comer, Interworking with TCP/IP: Principles, Protocols and Architecture Vol. 1, Prentice Hall, 5th Edition, 2006.
Proceedings of the XI International Conference ETAI 2013, 26th -28th of September 2013, Ohrid, Republic of Macedonia