Feasibility study resource optimisation

Feasibility Study Resource Optimisation for 3GPP LTE - SC-FDMA Uplink


Single carrier frequency multiple access (SC-FDMA) is investigated in terms for improving the performance of the uplink transmissions assuring that as many users are served and fairness between them exists. The system is investigated from the channel-dependent scheduling point of view in order to achieve multi-user diversity, increase in data throughput, and frequency selective diversity. The two subcarrier mappings available for SC-FDMA are: Localized FDMA (LFDMA) and Interleaved FDMA (IFDMA). The optimal and sub-optimal allocation techniques used for channel-dependent scheduling (CDS) regarding of what method is used, Greedy allocation or adaptive modulation when channel state information with feedback delay is available, show that there can be an improvement in data throughput and user resource allocation, as studied previously.

1. Introduction

1.1 Motivation

The motivation behind LTE is the deployment of a network that would manage to fulfil the needs for mobile broadband of the customers and improve the very successful solution of HSPA that is widely used in all parts of the world now.

The need of a simplified and faster network, determined the evolution from HSPA to LTE, which is capable of much higher data rates, richer user experience, better cost/performance ratio and increased cell capacity. All of these could be successfully accomplished by using an all- IP flat architecture with spectrum efficiency guaranteed by the possibility of allocating different spectrum widths, not just one like it was used in previous implementations, and very easy integration with the existing networks and technologies.

Higher spectral efficiency and the modulation techniques used both on downlink and uplink allow increased data rates with peaks of up to 173 Mbps in the downlink and 58 Mbps in the uplink and coverage over a larger area.

According to 3GPP Organization, "As the optimum evolution of 3GPP and 3GPP2 mobile technologies, published results of LTE versus WiMAX benchmarking indicate that in normalized metrics, LTE in the downlink is about 20% better in spectrum efficiency and average user throughput and much better in cell-edge user throughput. In the uplink, LTE is 50% better in both average and cell edge performance."[6]

In addition, the highly reduced cost per MB and the possibility of reusing some of the equipment used in the previous 3G technologies are a big plus for LTE networks deployment.

1.2 Objectives

The objectives of this dissertation project are:

  • to study the LTE technology with an emphasis on the uplink modulation scheme
  • to find methods that can be used in optimizing the resource allocation in LTE from the channel dependent scheduling point of view for the uplink SC-FDMA modulation
  • simulate this methods
  • show the improvements observed and compare them with the current developments

1.3 Aim

The aim of the dissertation is to determine an optimal channel assignment and power allocation scheme to maintain a minimum required data rate, because at a given time, one channel can be very good for one user and in the same time very poor for another one. This will be done by analysing the channel-dependent scheduling methods and finding the optimal method in which they can be used in order to assure the best performance and fairness between the users sharing the network. The system's performance using this optimal channel assignment will be evaluated through computer simulation.

1.4 LTE - the next natural step

With mobile broadband being a present reality and with the need of higher and higher data rates mobile operators must deploy an even better and faster network than the one used in present. The need of an even faster, efficient and highly reliable mobile network that will transform the way that users access, use, and experience this new mobile broadband solution and will support the deployment of a wide range of new and old services at a new level. In addition, interaction with the content will be changed and as some people call it, it can be the internet of things. LTE will not be just a new interface like some may consider, it is wanted to be a complete new ecosystem to meet the needs of the customers in the present and near future.

LTE is wanted to assure a smaller cost of data services at higher rates than in current HSDPA/HSUPA networks with a better use of spectrum and reduced latency.

1.5 Background for LTE

The continuous increase of data communications and the increasing demand of higher and higher bandwidth, the success of 3G technology that currently brings constant growth in number of subscriptions and in terms of revenues make the idea that a new and even faster technology must be deployed.

Taking the developments in 3G and the reliable spectral efficiency and mobility, LTE is developed with the purpose of assuring the high demand of data necessary in urban areas and not only. [1]

The new spectral configuration of LTE made possible to achieve even four times more spectral efficiency than in 3G/HSPDA networks which means a fantastic increase in data capacity in the same amount of spectrum as in 3G networks.

Besides the increased spectral efficiency, other major advantages and improvements like reduced network complexity, which means exponentially lower costs make this technology so amazing. Also in the same time lower costs from the operator side and increased data capacity gives more richness from the application side and cheap in terms of costs for the end customer. The new and improved spectral efficiency will show it's effects in a large number of factors that translates in higher number of voice users that can use the same cell, more services, applications and an overall greater experience using the same amount of bandwidth as in old 3G network. Another key characteristic of LTE's spectrum is flexibility. The fixed 5 MHz fixed channel bandwidth is substituted by channels with variable bandwidths from 1.25 to 20 MHz.

Taking into consideration the operator costs for the implementation of the new technology, these are reduced to a minimum because LTE has its backbone on the successful 3G technology. The mobile operators do not need to change their sites and can use the ones situated now in the current base stations.

LTE will be a complete packet based data network that will use IP and the Internet to carry all the data. This means that a lot of equipment that used to transport the information between the sites and the core network can be discarded. Being derived from an all-IP network also reduces the latency. The time it took for the information to get from one point to another in the network that used to be of 200ms now it is targeted at 5ms for small data packets.

The main requirements for the new technology are:

  • peak data rates of 100Mbps for downlink and 50Mbps for uplink
  • accommodation of minimum 200 users per cell with allocated channels of 5MHz
  • technologies used are OFDM, SC-FDMA, FDE, TDD, FDD, MIMO
  • mobility optimized for speeds between 0-15 km/h but can also assure high data rates at 120km/h and even supports speeds of 350 to 500km/h
  • adaptive modulation like QPSK for weak channels or 16QAM and 64QAM for strong channels
  • coding is made with convolutional codes or Rel-6 turbo code
  • supports MIMO multiplexing techniques 2x2, 2x4, 4x4
  • OFDMA multi-access scheme for download
  • SC-FDMA multi-access scheme for uplink

The different multi-access schemes chosen for download and upload are required for best performances in both cases. For the downlink in order to avoid the limiting capabilities of WCDMA used in previous 3G networks such as multipath interference OFDMA was chosen because it can support different bandwidths and is not fixed on a 5MHz. Also the smaller system complexity along with frequency diversity by spreading the carriers in the whole spectrum and time diversity are other advantages of this access scheme. It is also adaptable using the maximum multiplexing technique that a user can support. Some of the drawbacks of OFDMA are the high PAPR and the big influences from phase noise and offsets of frequency.

For the uplink the SC-FDMA is used because it provides a lower PAPR as OFDMA. It has a lot of similarities to OFDMA but is more suitable for the uplink because it requires lower power to transmit the same quantity of data. There is a different sub-carrier assigned to each frequency domain. Each user is assigned a number of sub-carriers in the frequency domain.

The ISI problem that appears in high bandwidth systems must be avoided and usually it is done with time delay equalization in the form of tap delay line filtering. To mitigate the frequency selective fading a multi-carrier techniques is required in order to divide the channel into smaller sub-bands. [4]

It is a well-known fact that at a moment in time it is possible for one channel to be very poor (deep fading) for one particular user but to have a fair or even good performance for another. In order to achieve am overall good performance and fairness among the users of the network a resource scheduler is very appropriate. A resource scheduler can assign appropriate available channels to a user in order to achieve the best performance possible at that particular moment. The resource scheduler is responsible for exploiting the diversity and frequency selectivity in order to increase the transmission rate available for each user by assigning the best available channel for the respective user.

2. Literature review

2.1 SC-FDMA background information

"SC-FDMA can be regarded as DFT-spread orthogonal frequency division multiple access (OFDMA), where time domain data symbols are transformed to frequency domain by DFT before going through OFDMA modulation [4]. The orthogonality of the users stems from the fact that each user occupies different subcarriers in the frequency domain, similar to the case of OFDMA. Because the overall transmit signal is a single carrier signal, PAPR is inherently low compared to the case of OFDMA which produces a multicarrier signal [5]."[4]

As stated by Farooq Khan in [11], for an uplink transmission, the transmission suffers of inter-symbol interference because the Walsh codes used for transmission are not orthogonal due to the lack of synchronization of the received codes at the base station. This translates into users interferencing one with each other and each of them contributing to the rise of the overall noise sent to the other users. In order for this noise not to increase, it is regulated at the base station and kept below a certain limit. This limit is called rise-over-thermal RoT and has the formula:

A SC-FDMA transmission has a series of different steps that must be accomplished in order to ensure correct communication. Firstly, the symbols are mapped into blocks of N symbols. Following this, an N-point DFT is performed in order to shift in the frequency domain. Each of these blocks are then mapped on different subcarriers to be transmitted. If M is the number of subcarriers, they are orthogonal between them and this avoids any possible co-channel interference. The output is then transformed back into time domain by performing an M-point IDFT. To ensure an accurate transmission a CP (cyclic prefix) is added, which is a copy of the last part from the data block and is put in front of the intended data block. The main role of CP is to provide a guard time and is greater than the maximum delay spread. CP also enables the discrete time circular convolution to be performed. At the receiver side, the time domain is converted into frequency domain in order for the carriers to be de-mapped and then decisions about the possible sent symbols is made with the help of MMSE and FDE.

2.2 Channel Dependence Scheduling (CDS)

In radio transmissions the channel gains on which the transmission takes place can be in deep fade for some users and for other users can have a good channel gain. In order for the channels to be correctly assigned, the base station has a scheduler which has the main property of assigning the user with the most appropriate channel for him.

The sub-carrier mapping and power allocation is a two step procedure that assigns the users with the best channel gains available for them and then the power assigned to the chunk of sub-carriers is split between them according to the water-filling policy.

An optimal allocation technique is complex and hard to implement but one low complexity channel dependent scheduling technique that offers a sub-optimal algorithm uses the greedy allocation.

The target of this technique is to assure a higher data rate with a limited transmit power that the mobile device can provide. The transmitted bandwidth B is split into multiple sub-carriers that are multiplexed in time and frequency. Each user can be allocated one or more sub-carriers from the total number, of sub-carriers available. The total number of carriers are split into a number of chunks. The appropriate number of chunks for each user and for a specific transmission is allocated by the channel dependent scheduler contained in the base station. This allocation is decided after estimation of the channel is made by sending trial symbols to evaluate the channel quality for that specific user. After an appropriate channel is chosen depending how good it is, has high or low gain, a specific data modulation technique is chosen to fulfil the transmission.

In order to assure a good quality of service the chunks are allocated according to greedy method and bit loading but, unfortunately the throughput increase would not be enough increased in order to be taken into consideration for such a high complexity of the system. Because of this, a sub-optimal chunk allocation is used.

The SC-FDMA is affected by inter-symbol interference and this is mostly removed by the use of minimum mean square error and frequency domain equalization.

For IFDMA chunk allocation the principle is as follows:

  • With the CNR calculated with the formula given previously the best users are added into a list of available users and also the available chunks are added.
  • Each user is allocated a corresponding number of chunks that is proportional to the ratio of the user's SNR to the sum of SNRs for all users.
  • The user that has been allocated the specific number of chunks for him is deleted from the list of available users and the same thing is done with the chunks
  • The process repeats until all the remaining chunks are allocated
  • Taking into consideration this approach, Junsung Lim, Hyung G. Myung, Kyungjin Oh, and David J. Goodman, observed that for moderate transmission rates the IFDMA gives a better performance due to lower PAPR and lower outage probability and for higher data rates LFDMA is appropriate by using power control to establish a power margin.[5]

    Another method used in channel dependent scheduling is using adaptive modulation and coding in order to achieve better performance in frequency selective fading channels and the use of a feedback sent through the channel. The channel dependent scheduler assigns the subcarriers according to the principle, each user takes the subcarrier that gives the best frequency response for him. This frequency response is known by estimating the channel first by sending trial symbols and the channel state information is being fed back for decision making at the base station. Unfortunately, not always all the information is correct because of channel estimation errors and the delay with which the feedback is returned. It is possible that because of the delay of the feedback the channel state to have changed when the channel is fast fading and this could have major effects in the correct estimation of the channel and on the system performance. This change is shown by the correclation coefficient:

    For an uplink transmission over SC-FDMA for a system with adaptive modulation at the channel dependent scheduler in the base station it has been observed that in order to eliminate the effects of feedback delay that can induce wrong estimation of channels affecting the user allocation and the modulation technique chosen it is highly recommended to introduce a delay between the channel estimation moment and the actual time of data transmission from the mobile terminals.

    By applying the adaptive modulation and CDS it has been observed that for LFDMA performes better in the case of imperfect channel state information but it is very sensible if the channel is changing fast.[7]

    In the paper, "Proportional Fair Scheduling of Uplink Single-Carrier FDMA Systems" by Junsung Lim, Hyung G. Myung, Kyungjin Oh and David J. Goodman the resource allocation of channels for the uplink SC-FDMA is described using two utility functions:

    • user data rate used for maximizing the systems capacity
    • logarithmic user data rate for proportional fairness

    These functions are proposed for managing the two available resources that can help us optimize and achieve better performances, time and frequency.

    In this paper the approaches mentioned earlier are revised in order to find an optimal sub-carrier allocation algorithm. This is done by using the greedy algorithm and assuming that the base station is aware which channel has the best performance for each user.

    The utility based equal-bit-equal-power chunk allocation scheduling is used for determination of an optimum assignment of channels. This can be done for both Localized FDMA and Interleaved FDMA. [8]

    1. Localized FDMA

    The Greedy chunk allocation algorithm stated above is based on marginal utility and has the following steps:

    • the chunk with the highest marginal utility is searched through all the available chunks and users
    • the algorithm searches for the user that can maximize the marginal utility and allocate the chunk that user
    • delete the chunk form the list with available chunks for assigning

    2. Interleaved FDMA

    For Interleaved FDMA the main task is also allocating a chunk to the user that can obtain the highest marginal utility. This process can be performed using two methods:

    1. Greedy allocation
    2. Flooring

    The Greedy allocation for Interleaved FDMA allocates an extra number of chunks to users even if they have already chunks allocated until all the remaining chunks will be assigned to the available channels at that moment.

    Flooring takes in consideration besides the Greedy algorithm the fact that in order to achieve an as low as possible PAPR and to keep the sub-carrier mapping equidistant the number of sub-carriers should be the power of 2. The maximum number of allocated chunks can be the closest integer, which is the power of two, but smaller, that gives us the number of chunks allocated with the Greedy algorithm. This translates into a tree structure as shown above:

    3. Methodology

    In order to achieve the proposed goals the methodology of the dissertation comprises the study of LTE technology especially the SC-FDMA uplink.

    The steps to fulfil the task are:

    • The first stage will be to acquire sufficient information for a good understanding of the technology and the current existing developments
    • The next step required is the simulation based on the basic SC-FDMA technique and developments in order to achieve the current known performances. This will be the starting point for optimization and achievement of better performance.
    • The second stage will be the design and implementation of an improved SC-FDMA in order to optimize the uplink transmission performances
    • The third step will contain a comparison between the current results and the improvements provided by the newly implemented design
    • The forth step will contain simulations regarding various scenarios to see how the optimized version behaves under different conditions

    3.1 Computer Simulation

    The project results will be obtained with the help of computer simulation software. In this case MATLAB Simulation Software will be used with the following packages:

    • Communications Toolbox 4.1
    • Signal Processing Toolbox 6.7
    • Statistics toolbox
    • Image Toolbox
    • Mapping Toolbox

    The use of simulation software is the most appropriate approach in our case because it would be highly inefficient to implement everything hardware just in order to see if it works and what performances we can get. Also, if there are problems with the design it is harder to identify the problem, where the problem comes from (hardware or software implementation). Another advantage of computer simulation is the reduced cost implied, and the rapidity with which we have a valuable result for the new design.

    The improvement of resource allocation will be simulated for various scenarios in order to see the impact for the uplink transmission in different possible situations that can be encountered in day by day use.

    The simulation method used will be Monte Carlo, which is the most appropriate method in such cases because it provides a true estimate of the error probability when the number of simulations is very large. The Monte Carlo Simulation method gives an approximate value for the searched outcomes after running the simulation a large number of times.

    The use of the Monte-Carlo simulation is appropriate for the study in this case because the performance of the implemented optimization method will be tested for different scenarios that each of them requires and the performance simulation requires many runs.

    In the first stage, the SC-FDMA will be simulated in order to achieve the current performances and after the implementation of the optimized version for the uplink it will be compared with the current achievements.

    In the second stage of simulation part of the dissertation, the results of the Monte-Carlo technique will be compared with the initial results in order to see the improvements introduced by the new implementation for different scenarios and to check if its influence is better just for some cases of usage or for all. In case there is time for the situations where the algorithm has poorer performaces an improvement will be attempted.

    4. Project planning and milestones

    4.1 Matlab Environment Configuration and Learning

    In the first phase Matlab Software will be installed and a part of the packages that it offers regarding mobile wireless communications. The software programs and packages used include the following:

    • Matlab 7.8.0 (R2009a)
    • Communications Toolbox 4.1
    • Signal Processing Toolbox 6.7
    • Statistics toolbox
    • Image Toolbox
    • Mapping Toolbox

    Familiarization with Matlab programming and the packages used is needed in this primary part.

    4.2 Programming of the SC-FDMA uplink transmission

    After the theoretical knowledge gained during reading of the literature review and the information regarding the current performances of the uplink the transmission will be implemented according to the SC-FDMA standard and checked for performances in order to achieve the current performances for the present developments. This will be the start point of the new improved implementation.

    The current methods for uplink transmission will be analized in the following manner:

    • Analyze the current proposed models for SC-FDMA
    • Simulate the performances with the use of Matlab

    4.3 Compare simulated results with current developments

    The implemented uplink is simulated and the results are compared with the results of others in order to check the accuracy of the implementation and to be sure that the starting point of the improved version is correct.

    This is done in order to see that the results of the implementation described above are the same with the results stated in literature and in current proposed models.

    4.4 Design optimization solution

    At this step the implementation of the improved solution based on the present developments is trying to be achieved. This will be done by modifying and varying different factors that can influence the performance of transmission. Some of these factors might be:

    • Roll off factor
    • Number of sub-carriers used
    • Number of data blocks sent
    • arrier frequency offset
    • Bandwidth expansion factor
    • Etc...

    4.5 Simulate the designed solution

    The optimized LTE uplink implementation is simulated and the results are recorded for different scenarios.

    Using different scenarios will give a better overview in the performance improvement and if the improvement can be seen in more than one scenario possible in real life.

    4.6 Compare results with current developments

    The results obtained previously are compared with the other current developments to see how big are the improvements with the designed methods and in which cases we obtain a better or a worse result.

    4.7 Improvement of results

    Check if the current designed implementation is possible to be improved overall or only in the parts where are obtained worse results than in the current developments.

    4.8 Dissertation writing

    The results obtained until this point are written in the dissertation format according to the standards imposed by the university.

    4.9 Gantt Chart

    5. References

    [1] Technical overview of 3GPP LTE, Hyung M. Myung, 2008

    [2] Long Term Evolution, Antonis Hontzeas, from< http://considerations.wordpress.com > accessed at 15.05.2010

    [3] LTE for 4G Mobile Broadband, Farooq Khan, Cambridge University Press 2009,

    [4] Intoduction to Single Carrier FDMA, Hyung M. Myung, EURASIP 2007

    [5] "Channel-Dependent Scheduling of Uplink Single Carrier FDMA Systems", Junsung Lim, Hyung G. Myung, Kyungjin Oh, David J Goodman, IEEE, 2006

    [6] Long Term Evolution - White paper, Nokia Siemens Networks Corporation, B301-00342-EF-200812-1-EN

    [7] "Channel-Dependent Scheduling of an Uplink SC-FDMA System with Imperfect Channel Information", Hyung G. Myung*, Kyungjin Oh, Junsung Lim, and David J. Goodman

    [8] "Proportional Fair Scheduling of Uplink Single-Carrier FDMA Systems" by Junsung Lim, Hyung G. Myung, Kyungjin Oh and David J. Goodman

    [9] B. Sklar, "Rayleigh Fading Channels in Mobile Digital Communication Systems: Part I: Characterization," IEEE Commun. Mag., vol. 35, no. 7, Jul. 1997, pp. 90-100.

    [10] Single carrier FDMA : a new air interface for long term evolution, Hyung G. Myung, David J. Goodman, Wiley, 2008, ISBN 978-0-470-72449-1

    [11] LTE for 4G Mobile Broadband - Air Interface Technologies and Performance Farooq Khan, Cambridge University Press 2009

    [12] Low-Complexity Iterative Receiver for Interleaved FDMA (IFDMA) with Cyclic Delay Diversity, Jiang Xiang, Yu Cai, Ying-Chang Liang, Kwok Hung Li and Kah Chan Teh

Please be aware that the free essay that you were just reading was not written by us. This essay, and all of the others available to view on the website, were provided to us by students in exchange for services that we offer. This relationship helps our students to get an even better deal while also contributing to the biggest free essay resource in the UK!