An (b) the home subscriber server (HSS)

An LTE network comprises two main entities known as the
evolved universal terrestrial radio access network (EUTRAN) and the evolved
packet core (EPC). The E-UTRAN includes the eNB which interact among themselves
via an x2 interface and with the UEs over the air over the Uu interface. In
addition, E-UTRAN communicates with the EPC via the S1 (i.e. S1MME/S1U)
interface and is also in charge of modulation/demodulation, Channel
coding/decoding, radio resource control, radio mobility management, and
detection and correction of errors in the transmitted data. On the other hand,
the EPC comprise the following entities: (a) the mobility management entity
(MME) handles security procedures, sessions between UE and network, and
location management. It connects to the eNB via the S1MME interface; (b) the
home subscriber server (HSS) connects to the MME via the S6a interface. It
hosts the home location register and the authentication center; (c) the serving
gateway (SGW) serves as local mobility manager anchor within the E-UTRAN or
mobility with other 3GPP technologies. It connects to the eNB, MME, and the
serving GPRS support node (SGSN) through the S1U, S11 and S4 interfaces,
respectively; (d) the packet data network gateway (PDN) routes the traf?c to
the internet via the SGi interface. It also connects to the SGW , the policy
charging and rules function server (PCRF) and the evolved packet data gateway
(ePDG) via interfaces S5/S8, S7 (Gx) and S2b, respectively. (e) the PCRF
handles the service policy, charging information and QoS parameter information
for each user session. (f)?nally, the ePDG interconnects the EPC to untrusted
non-3GPP networks that need a secured access 26.

B. LTE channels

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

The data to be transmitted over an LTE network is classi?ed
into user plane and control plane data. The former is actual data intended for
the user while the latter is the necessary data for a successful delivery of
the user plane data. Ef?cient classi?cation of these data is required to
identify its type and purpose. There are three types of channels in LTE known
the logical, transport and physical channels. These channels are respectively
implemented by the radio link control (RLC), the medium access control (MAC)
and the physical layer (PHY). The RLC performs the logical categorization of
the control plane data into ?ve logical channels as follows: (a) the paging
control channel (PCCH) which carries the paging message; (b) the broadcast
control channel (BCCH) which contains the system critical information such as
bandwidth, reference signal power, antenna con?guration…; (c) the common
control channel (CCCH) which is used by the UE to acquire control information
when no radio connection is established; (d) the downlink control channel
(DCCH) is used to enable the UE and the eNB to exchange control information on
a dedicated bearer resource; (e) the multicast control channel (MCCH) carries
control information of a group of UEs in a cell for multicast/broadcast
services. The RLC also classi?es the user

7

plane data in two logical channel as: (a) the dedicated
traf?c channel (DTCH) which carries the user plane data in both uplink and
downlink directions, and (b) the multicast traf?c channel (MTCH) which is a unidirectional
channel used to transmit the user data from the eNB to multiple UEs. The RLC
passes the categorized data to the MAC layer where the transport channels are
assigned. In the downlink, there are four transport channels. The PCCH and BCCH
are mapped to the paging channel (PCH) and the broadcast channel (BCH) while
the CCCH, DCCH, MCCH and DTCH are multiplexed into the downlink shared channel
(DL-SCH). Since the BCCH carries various types of signaling data, the system
information is split into: (a) the master information block (MIB) which carrier
system critical information to acquire a cell is carried by BCH, and (b) the
system information block (SIB) which contains dynamic system information to
insure a reliable radio connection in the uplink is mapped to DL-SCH. As a
result, decoding the SIB is important for the user. Yet, since its location on
the physical resource grid is unknown, the user is required to decode MIB ?rst
to get the system critical information and the scheduling information for SIB.
In the uplink, the transport channels used are similar to those in the
downlink. The logical channels such as CCCH, DCCH and DTCH as mapped to the
uplink shared channel (ULSCH). Since the user does not send paging,
broadcast/multicast messages in the uplink, these channels do not exist in the
uplink. However, there is one transport channel known as random access channel
(RACH) which is used at the initial stage when the UE is not synchronized with
the eNB in the uplink. It is the only channel which does not have a
corresponding logical channel. In addition to multiplexing and demultiplexing
the logical channels, the MAC scheduler also handles: (a) hybrid automatic
repeat request (HARQ) in case the received data fails the cyclic redundancy
check (CRC) or cannot be decoded, (b) random access process, and (c) QoS class
identi?er (QCI) function through logical channel prioritization. This where the
MAC layer decides the amount of data from each logical channel to be included
in the MAC packet data unit (PDU)/transport block (TB). The TB size is based on
the uplink resource grant request message. Since the TB size is ?xed (e.g.
10KB) ?lling it in the order of the priority of the logical channels means that
higher priority data would possibly ?ll all the TB leaving the lower priority
data out. This is known as channel starvation. To avoid this, one priority
bitrate (PBR) is assigned to each logical channel. The TB is then ?lled based
on the PBR. (d) decision regarding the appropriate modulation and coding scheme
(MCS) is also made. The MAC scheduler receives the channel state information
(CSI) in form of channel quality indicator (CQI) based on which it decides the
MCS to be used for the communication. The data in the DL-SCH contains the user
plane traf?c which needs to be delivered irrespective of the channel quality.
Hence, higher MCS such as 64QAM (quadrature amplitude modulation) is used when
the channel is good. Otherwise lower MCS such as quadrature phase shift keying
(QPSK) is used. Since the latter is the most robust

modulation scheme used in LTE, it is used for signaling
traf?c. The MAC scheduler also handles the three following functions
implemented in the physical layer such as deciding the number of terminals to
transmit, determining the set of resource blocks for the UEs in the downlink,
and the selecting the TB size. The transport channels such as PCH and DL-SCH
are mapped to the physical downlink shared channel(PDSCH). On the other hand,
BCH and MCH are mapped to the physical uplink shared channel (PU-SCH) while the
physical RACH is used for RACH. When a UE receives physical data from the eNB
it needs to locate the DL-SCH and determine the MCS to used. Hence, level one
(L1) and level two (L2) control signals are used. The physical downlink control
channel carries the downlink control information (DCI) about the user data, the
SIB and the uplink power control information and HARQ control information. The
physical control format indicator channel (PCFICH) carries the control format
indicator (CFI) which tells the number of OFDM symbols used for PDCCH. Next,
the physical hybrid ARQ indicator channel (PHICH) carried the hybrid ARQ
indicator (HI) which comprises the HARQ and the ACK/NACH messages to check for
successful data delivery. In the uplink, the physical uplink control channel
(PUCCH) carried the uplink control information (UCI) for the user traf?c and is
mapped to UL-SCH. On the other hand, the physical uplink channel (PUCCH)
carries the CQI, ACK/NACH and the scheduling request messages to the eNB. At the
physical layer, each TB is subject to thorough digital signal processing prior
to being sent over the air. Large TBs (e.g. > 10KB) are fragmented into
smaller block with a CRC attached to each. Next, channel coding is performed
using a Turbo coding technique which convert each bit of data into 3bits. The
resulting data stream undergoes rate matching and scrambling before being
mapped to the correct modulation as chosen by the MAC scheduler. After
precoding, the bits are ?nally mapped to the REs and endergo the orthogonal
frequency division multiplexing OFDM process before being transmitted over the
Uu interface.

C. LTE data transmission

The data transmission in LTE is driven by OFDM techniques.
It is performed over parallel subcarriers of 15KHz, each subdivided into time
slots as blocks of one symbol of 66.7µs also known as resource elements (RE).
Transmission of data over these REs requires modulating the data using an
appropriate modulation scheme. Such scheme depends on the physical channels
mapped on the LTE resource grid. Suppose data, such as video generated at a
given bitrate, is to be transmitted using M ? aryQAM. The data is split into M
parallel streams of log2M bits. The assignment is performed based on the M
?aryQAM constellation where the amplitude and phase of the subcarrier is
adjusted with respect to those of the data stream before the data can be placed
on the REs. Therefore the modulated data is product of the complex modulation
symbol by the corresponding subcarrier frequency. For instance, in a 20MHz
bandwidth, there are 100 resource blocks (RB) where each contains 12
subcarriers.

8

The Inverse Fast Fourier Transform (IFFT) is used at the
transmitter to convert the parallel frequency domaine signals into samples of
composite time domain signals. The IFFT samples must be taken above the Nyquist
rate. In the 20MHz bandwidth, 2MHz is reserved from guard band. So, the highest
frequency components range from 9MHz to -9MHz. Thus, the sampling rate must be
at least 18Mega Sample per second. Since the transmitted data may experience
degradation due to intersymbol interference (ISI) and delay spread as a result
of multipath propagation, even though the effect can be reduced by guard
periods, the sudden ?uctuation in the time domain during the transition from
symbols to guard periods can also disrupt the orthogonality between
subcarriers. Therefore, a cyclic pre?x (CP) is added to each transmission cycle
by taking IFFT samples from the end of a period and placing them at the
beginning. With that said, the samples signal is converted into a wave form by
using a digital to analog (A/D) converter and modulated at the desired radio
frequency. OFDM is used in LTE in the downlink, where a each symbol is mapped
to one subcarrier. But doing this also results into a high peak to average
power ratio (PAPR) as a result of the IFFT summation of multiple parallel
subcarriers. Since the UE has a limited power capability, single carrier
frequency division multiple access (SC-FDMA) is used in LTE in the uplink,
where the symbol duration per subcarrier is shorter than in OFDMA given that
each symbol is split across all the subcarriers in each RB.

 

Prior to sending traf?c, each UE needs to get uplink
synchronization and also get a radio resource allocated in the uplink direction.
Such need arises when the UE needs to register on the LTE network, initiate
data transaction in the downlink or uplink directions if the UE falls in a
nonsynchronized status, handover or timing advance request. The UE gets the
SIB-2 parameters from the eNB in order to initiate a RACH procedure by sending
a RACH preamble to the eNB. A preamble comprises a CP and a sequence derived
from the Zadd-off chu scheme. This makes the preamble generated by all UEs
orthogonal to each other. There are 64 preambles, some of which are reserved
for use by the eNB in case for non contention based RACH procedures such as
handover. Each UE chooses from the remaining preambles for contention based
RACH procedures such as synchronization for normal data download/upload. The
available preamble to the UEs are grouped in two categories where the size of
each is determines by the number of preambles that it contains. The UE
determines the format and timing of the preamble through the con?guration index
parameter contained in SIB-2 and selects a preamble from one of the preamble
groups based on the size of layer two (L2) or layer three (L3) message to be
transmitted to the eNB. In the frequency domain, each UE uses six physical RBs
for its RACH request message. Since the UE does not send any identi?er when
sending the preamble, the eNB calculates the UE identi?er as random access
radio network temporary identi?er (RA-RNTI) based

on the reception time slot of the preamble. The eNB also
computes the temporary cell RNTI (C-RNTI) to be used from further
communications with the UE. In the RACH response (RAR) message sent by the eNB
to the UE, parameters necessary for the UE to transmit L2/L3 messages are
included. Such parameters comprise the timing advance (TA) number of RBs to be
used by the UE for uplink transmission, the MCS index to be used, the hopping
?ag (FH), the power to be used on PU-SCH, the channel quality indicator (CQI)
?eld… For the purpose of this study, we limit the number I of UEs to 60 and
also propose that the RAR messages includes an information regarding the SA/QoE
coef?cient ?elds used by the UE to also report its SA coef?cients ?, ?, and ?,
as well as QoE coef?cients a, b and c. In our experiment, a single eNB receives
individual video streams from 60 UEs in the uplink direction over a Bandwidth
(B) of 20MHz. Each UE i transmits a video content is at a bitrate bi r ? bi
rmax , where bi rmax represents the maximum achievable bitrate for UE i. The
values SA coef?cients for each UE change with respect to the radio link quality
for that UE. Since B = 20MHz, a total of 100RBs are available for all the users
within that cell. Each RB comprises 12 subcarriers and 7 symbols (assuming a
regular CP size), meaning that 8400 REs are available for all the UEs within a
time slot of 0.5ms. In excellent channel condition, up to 64QAM could be used.
Hence 6 bits are carried by each RE. Assuming that 2×2 MIMO is used, we get a
total throughput of 201.6 Mbps at the physical layer. But the overhead is
removed as follows: (a) 6% for cyclic pre?x, (b) 10% for guard band, (c) 10%
for signaling overhead, and (d) 14% for pilot overhead. This limits the maximum
achievable throughput to 120Mbps. Therefore, in our experiment, we limit the
aggregate bitrate to Br = 120Mpbs for all the UEs. Once each UE receives a RAR
on the DL-SCH, it saves the C-RNTI, applies the timing correction to get
synchronized in the uplink and proceed with the RRC connection request
procedure until it receives an RRC connection setup con?rmation message from
the eNB. Only then, UEs start streaming video to eNB which uses the information
in addition the CQI in order to decide of the modulation scheme for each UE.
The eNB acts as a centralized SA optimizer which account for the feedback
received from the individual UEs and allocates resources accordingly. However,
before discussing our approach, we provide a brief background on two
optimization techniques known as branch and bound and liner programming.