5G Korea -

Network Simulator v1.0

User Manual
Jan 2019

Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea,

Chungnam National Univ., Daejeon, Korea,
Hanyang Univ.(ERICA), Ansan, Gyeonggi, Korea,
Hankyong National Univ., Anseong, Gyeonggi, Korea,
Korea Univ., Seoul, Korea,
Dongguk Univ., Seoul, Korea,
Seoul National Univ., Seoul, Korea,
Yonsei Univ. Seoul, Korea,

Web: http://5gopenplatform.org

1
Abstract

This document contains an overall description of 5G K-NetSim (5G Korea-Network Simulator) and how
to use it for users. There is a list of features of 5G K-NetSim, a structure, and parameter descriptions in
this document. 5G K-NetSim has been developed with a modular and flexible architecture, users can
easily adjust the simulator to evaluate their algorithms or functions. This simulator is released under an
academic, non-commercial use license.

2
Contents
1. Introduction ·········································································································· 7
1.1 Purpose ········································································································· 7
1.2 License ········································································································· 7
1.3 Contact points ································································································· 7
1.4 Development environment ·················································································· 8
1.5 Download ······································································································ 8
1.6 Definition, symbols and abbreviations ···································································· 8
1.6.1 Definition ································································································· 8
1.6.2 Symbols ··································································································· 8
1.6.3 Abbreviations ····························································································· 8
2. List of features ·····································································································10
2.1 Performance metrics ·························································································10
2.2 Simulation parameters for configuration ·································································10
2.2.1 General parameters ·····················································································10
2.2.2 5G NR NSA ·····························································································11
2.2.3 5G NR Channel ·························································································11
2.2.4 5G Core ··································································································12
3. Network Simulator Structure·····················································································13
3.1 Overview ······································································································13
3.2 Introduction to ns-3 ··························································································13
3.3 Directory structure ···························································································14
3.4 Simulation methodology ····················································································14
3.5 Main components in 5G K-SimNet ·······································································15
3.5.1 Dual radio protocol architecture ······································································16
3.5.2 NR radio protocol ·······················································································17
3.5.3 SDN/NFV ································································································17
3.5.4 5G Core
4. Parameter description ·····························································································18
4.1 General parameters ··························································································18
4.2 5G NR NSA ··································································································19
4.3 5G NR SA ····································································································20
4.4 SDN/NFV ·····································································································23
5. Test scenarios ······································································································26
5.1 Overview ······································································································26
5.2 Parameter setting ·····························································································27
5.3 Results ·········································································································29

3
6. How to use ··········································································································31
6.1 How to install 5G NR NSA simulator ····································································31
6.2 Workspace ····································································································31
6.3 How to install 5G SDN/NFV simulator ··································································32
6.4 How to run your own simulation ··········································································34
7. Change Log ·········································································································36
8. References ··········································································································37

4
List of Figures

Figure 1 Directory structure ···················································································14

Figure 2 Simulator procedures ················································································14
Figure 3 Simulator architecture of 5G K-SimNet ··························································15
Figure 4 User plane diagram of LTE-NR dual connectivity in downlink case ·························16
Figure 5 SDN/NFV simulation process ······································································18
Figure 6 Code block diagram ··················································································25
Figure 7 Simulation results of LTE-NR dual connectivity ················································30
Figure 8 Mobility performance in LTE-NR dual connectivity ···········································30
Figure 9 User RTT in light load condition and heavy load condition ···································31
Figure 10 virt-5gc-vminput.input format ····································································35
Figure 11 virt-5gc-toposample3.input format ·······························································36

5
List of Tables

Table 1 Definition of frequency ranges ······································································20

Table 2 NR operating bands in FR1 ··········································································20
Table 3 Number of OFDM symbols per slot, slots per frame, and slots per subframe for normal ··21
Table 4 Maximum transmission bandwidth configuration NRB ·········································22
Table 5 Channel Bandwidths for Each NR band ···························································22
Table 6 Slot formats·····························································································23
Table 7 Simulation parameter settings ·······································································27

6
1. Introduction
This document describes 5G network simulator. 5G-K network simulator is implemented based on ns-3
network simulator. It is hard to develop the new network simulator because network simulator should
contain all the protocol stacks with related to end-to-end communication. Even if we develop the new
network simulator, it is difficult to prove the reliability of our network simulator because of massive
characteristics of network simulator. Hence, we develop the 5G-K network simulator based on ns-3
simulator. ns-3 network simulator is one of representative network simulators in communication and
computer network research. ns-3 models network entities and structs complete set of the network protocol
stack, so it can predict network behaviors more accurately. However, conventional ns-3 simulator lacks of
consideration about virtualization effects or cutting edge of SDN technologies, or support of dual
connectivity of LTE and 5G networks. For simulating 5G network system, it is necessary to see virtualized
effects or SDN functions, or test dual connectivity scenario for deployment of 5G network. Our 5G-K
network simulator includes OpenFlow controller, OpenFlow switches, modules for evaluating
virtualization effects and dual connectivity of LTE and 5G network. Users who want to simulator 5G
network system can evaluate their own network system topology with our 5G-K network simulator.

1.1 Purpose
This document is intended for use by software engineers working directly on the network simulator. In
the paper we explain about two main components of the simulator (DC and SDN/NFV), parameters and
metrics of the simulator, design considerations, simulator architecture, and how to use the simulator.

1.2 License
TBD

1.3 Contact points

Jintaek Lim

Korea Advanced Institute of Science and Technology (KAIST)

335 Gwahangno, Yuseong-gu, Daejeon, 305-701, Republic of Korea

Office: Room 717, IT Convergence Building (N1)

Tel: +82-42-350-5472

E-mail: jtyim@kaist.ac.kr

7
1.4 Development environment
 OS: Linux Ubuntu 14.04
 Language: C++ (gnuplot for GUI, some python codes in ns-3)
 Compiler: Waf (python-based build system)
 Setting: ns-3 version 3.26 based simulator (pre-configuration is required. Please refer to
https://www.nsnam.org/docs/release/3.26/tutorial/html/getting-started.html)

1.5 Download
Download source code from the 5G web site(http://5gopenplatform.org)

1.6 Definition, symbols, and abbreviations

1.6.1 Definition
Word Definitions

1.6.2 Symbol
Symbols Descriptions

1.6.3 Abbreviations
Abbreviations Descriptions
5G NR 5th generation new radio
DC dual connectivity
NSA non-standalone
SA standalone
SDN software-defined network
NFV network function virtualization
8
9
2. List of features
2.1 Performance metrics
 User throughput

– User throughput is an important metric for network simulators, while it can show general
network performance for each user device.

 User round trip time

– User RTT can show timely change of packet transfer performance. If RTT increase/decrease
sharply in short period of time, there might be special reasons such as VNF scaling, and so
on.

 TCP congestion window size

– With TCP congestion window size, user can verify the operation of network simulator and
changes of RTT or throughput.

 Scaling delay

– Scaling delay means additional delay components compared to non-virtualized core

networks. Additional delays depending on where network operators put VNFs (VNF
topology) can be evaluated.

 SDN switch throughput per port

– With SDN controller and switches, user can engineer traffic flowing the simulation
networks. SDN switch throughput per port can show results of traffic engineering.

2.2 Simulation parameters for configuration

2.2.1 General parameters
 X2 interface settings (src/lte/helper$ point-to-point-epc-helper.cc )

– X2 link data rate (m_x2LinkDataRate)

– X2 link delay (m_x2LinkDelay)

– MTU of X2 link (m_x2LinkMtu)

 S1AP interface settings (src/lte/helper$ point-to-point-epc-helper.cc )

– S1AP link data rate (m_s1apLinkDataRate)

10
 – S1AP link delay (m_s1apLinkDelay)

– MTU of S1AP link (m_s1apLinkMtu)

 S1-U interface settings (src/lte/helper$ point-to-point-epc-helper.cc )

– S1-U link data rate (m_s1uLinkDataRate)

– S1-U link delay (m_s1uLinkDelay)

– MTU of S1-U link (m_s1uLinkMtu)

2.2.2 5G NR NSA (src/lte/model$ mc-enb-pdcp.cc)

 PDCP reordering settings (mc-ue-pdcp.cc)

– PDCP reordering enabler ( m_isEnableReordering)

– PDCP reordering timer (expiredTime)

 Type of splitting algorithms(m_isSplitting)

– Splitting to single eNB (0 or 1)

– Alternative splitting scheme to two eNB (2)

– Assistant info-based splitting scheme (3,4)

2.2.3 5G NR Channel
 Operating band parameters (src/mmwave/model$ mmwave-phy-mac-common.cc)

– Center frequency (m_centerFrequency)

 Channel parameters (src/mmwave/model$ mmwave-3gpp-propagation-loss-model.cc)

– Pathloss model (&MmWaveHelper::SetPathlossModel)

(~/GigaKoreaProject/mmwave_new_handover/ns3-mmwave/src/mmwave/helper$ mmwave-
helper.cc)

– Scenario (m_scenario)

– Channel condition (m_channelConditions)

11
 – Optional Nlos (m_optionNlosEnabled)

– Shadowing (m_shadowingEnabled)

 Frame structure and physical resources parameters (src/mmwave/model$ mmwave-phy-mac-

common.cc)

– Channel bandwidth (m_numRb)

– Symbols per slot (m_symbolsPerSlot)

– Slot length (m_slotsPerSubframe)

– The number of layer in gNB (m_numEnbLayers)

 gNB-side parameters (src/mmwave/model$ mmwave-enb-phy.cc)

– Tx power (m_txPower)

 UE-side parameters (src/mmwave/model$ mmwave-ue-phy.cc)

– TxPower (m_txPower)

 gNB antenna parameters (src/mmwave/model$ mmwave-enb-net-device.cc)

– The number of vertical antenna elements (m_vAntennaNum)

– The number of horizontal antenna elements (m_hAntennaNum)

– The number of polarization dimensions (m_polarNum)

– The number of vertical TXRUs (m_vTxruNum)

– The number of horizontal TXRUs (m_hTxruNum)

– Antenna connection mode (m_connectMode)

– Bearing angle (m_rotation.x)

– Downtilt angle (m_rotation.y)

– Slant angle (m_rotation.z)

– Polarization slant angle (0 or 45 degree) (m_pol)

12
 UE antenna parameters (src/mmwave/model$ mmwave-ue-net-device.cc)

– The number of vertical antenna elements (m_vAntennaNum)

– The number of horizontal antenna elements (m_hAntennaNum)

– The number of polarization dimensions (m_polarNum)

– The number of vertical TXRUs (m_vTxruNum)

– The number of horizontal TXRUs (m_hTxruNum)

– Antenna connection mode (m_connectMode)

– Bearing angle (m_rotation.x)

– Downtilt angle (m_rotation.y)

– Slant angle (m_rotation.z)

– Polarization slant angle (0 or 45 degree) (m_pol)

 MC UE antenna parameters (src/mmwave/model$ mc-ue-net-device.cc)

– The number of vertical antenna elements (m_vAntennaNum)

– The number of horizontal antenna elements (m_hAntennaNum)

– The number of polarization dimensions (m_polarNum)

– The number of vertical TXRUs (m_vTxruNum)

– The number of horizontal TXRUs (m_hTxruNum)

– Antenna connection mode (m_connectMode)

– Bearing angle (m_rotation.x)

– Downtilt angle (m_rotation.y)

– Slant angle (m_rotation.z)

– Polarization slant angle (0 or 45 degree) (m_pol)

13
2.2.4 5G Core
 AMF-related parameters (virt-5gc.cc, virt-5gc-node.cc, topology input file (user input))

– AMF cpu capacity (Virt5gcNode::cpuSize)

– AMF memory capacity (Virt5gcNode::memSize)

– AMF disk capacity (Virt5gcNode::diskSize)

– AMF bandwidth (Virt5gcNode::bwSize)

– AMF workloads (Virt5gcNode::cpuUtil, Virt5gcNode::memUtil, Virt5gcNode::diskUtil,

Virt5gcNode::bwUtil)

– AMF x axis (Virt5gcNode::node_x)

– AMF y axis (Virt5gcNode::node_y)

 SMF-related parameters (virt-5gc.cc, virt-5gc-node.cc, topology input file (user input))

– SMF cpu capacity (Virt5gcNode::cpuSize)

– SMF memory capacity (Virt5gcNode::memSize)

– SMF disk capacity (Virt5gcNode::diskSize)

– SMF bandwidth (Virt5gcNode::bwSize)

– SMF workloads (Virt5gcNode::cpuUtil, Virt5gcNode::memUtil, Virt5gcNode::diskUtil,

Virt5gcNode::bwUtil)

– SMF x axis (Virt5gcNode::node_x)

– SMF y axis (Virt5gcNode::node_y)

 UPF-related parameters (virt-5gc.cc, virt-5gc-node.cc, topology input file (user input))

– UPF cpu capacity (Virt5gcNode::cpuSize)

– UPF memory capacity (Virt5gcNode::memSize)

– UPF disk capacity (Virt5gcNode::diskSize)

14
 – UPF bandwidth (Virt5gcNode::bwSize)

– UPF workloads (Virt5gcNode::cpuUtil, Virt5gcNode::memUtil, Virt5gcNode::diskUtil,

Virt5gcNode::bwUtil)

– UPF x axis (Virt5gcNode::node_x)

– UPF y axis (Virt5gcNode::node_y)

 Virtualization parameters (virt-5gc.cc)

– Scaling policy (memory split ratio) (Virt5gc::scaleInRate, Virt5gc::scaleOutRate)

– VM provisioning delay (Virt5gc::allocDelay)

 Data center parameters (virt-5gc-node.cc, virt-5gc-vm.cc, topology input file (user input) )

– Data center topology

 Node ID (Virt5gcNode::node_id)

 Node ToR ID (Virt5gcVm::ToR)

 Node physical machine ID (Virt5gcVm::pm)

 Link bandwidth (Topology input file)

 Link edge nodes (Topology input file)

 Software-Defined Networking (SDN) parameters (User input file (scratch file), ovs-point-to-
point-epc-helper.cc, qos-controller.cc )

– Inter-switch data rate (ovs-point-to-point-epc-helper.cc (DataRate(“10Mbps”)) )

– gNB-Switch-GW data rate (ovs-point-to-point-epc-helper.cc (ovs-point-to-point-epc-

helper.cc (m_s1uLinkDataRate) )

– remote host-GW data rate (scratch file (DataRate) )

– In-port, out-port (qos-controller.cc (in_port, output) )

– Output port group (qos-controller.cc (group-mode) )

– QoS weight (qos-controller.cc (weight) )

3. Network Simulator Structure

15
3.1 Overview
Network simulator is developed based on ns-3 version 3.26 and therefore the structure inherits ns-
3s’ structure. In this chapter, we describe the introduction to ns-3, the simulation methodology,
directory structure, and so on.

3.2 Introduction to ns-3

ns-3 is a discrete-event network simulator for Internet systems, targeted primarily for research and
educational use. ns-3 is free software, licensed under the GNU GPLv2 license, and is publicly
available for research, development, and use. The conventional ns-3 provides the LTE radio protocol
in order to evaluate the performance of LTE standalone scenario. Researchers can evaluate the
performance of LTE-NR dual connectivity, which is promising technology in 5G era, by using 5G-K
network simulator, providing the LTE-NR dual radio protocol stack, packet sequencing/reordering
supporting split bearer, and packet split/routing algorithm enhancing the user data throughput. The
conventional ns-3 simulator also lacks of virtualization modules. Users cannot simulate scenarios of
virtualized core networks. 5G-K network simulator can provide various simulation results depending
on users’ inputs including VNF placing topology. On the other hand, ns-3 has OpenFlow 1.3 module,
however, it cannot work with cellular (LTE-EPC, NR, etc.) modules. 5G-K network simulator let users
be able to simulate a scenario that SDN and cellular modules are used simultaneously. You can find
more about ns-3 at https://www.nsnam.org.

3.3 Directory structure

5G K-SimNet
src
lte
scratch
mmwave

virt-5gc
...

Figure 1 Directory structure

All the source codes are implemented in ‘src’ folder. For example, LTE module can be found in
‘src/lte’. Each module has sub-modules, helper and model. Protocols, algorithms, etc. are implemented
in ‘model’. Helper makes the configuration and installing interfaces easier.

3.4 Simulation methodology

16
 Figure 2 Simulator procedures

Simulation will proceed as described in above figure, topology definition, model, configuration,
execution, and output analysis. In the first stage, topology definition, the nodes needed for experiment
and interfaces between nodes are defined. In model phase, protocol stacks, such as TCP, LTE and Wi-
Fi, and applications are installed in nodes. Attributes, environmental variables, and command line
arguments are configured in the following stage. The scratch code includes topology definition, model
and configuration. The simulation runs with command line execution in Linux OS. The experimental
results are logged in text format or specific format, e.g., Wireshark. You can study the result with
gnuplot, MATLAB, or Wireshark.

3.5 Main components in 5G K-SimNet

Figure 3 Simulator architecture of 5G K-SimNet

17
 Figure 3 represents the simulator architecture of 5G K-SimNet. The 5G K-SimNet is being
expanded from ns-3, and 5G K-SimNet currently supports LTE-NR (New Radio) dual connectivity
(DC) and 5G core module including SDN/NFV (Software-Defined Network/Network Function
Virtualization) for end to end performance.
LTE-NR dual connectivity module supports one of 5G's main features to compensate for the
instability of NR link. To support dual connectivity, 5G K-SimNet remodified the existing LTE and
NR modules to develop a new LTE-NR dual module. Many layer layers have changed, but NAS,
PDCP, RRC, and RLC layers have been modified. Through the LTE-NR dual module, incoming
packets through the eNB are distributed by X2-inteface and are transmitted to the UE, are reordering.
5G core network functions consist of AMF, SMF and UPF. These functions are essential
components for simulating 5G data transmission. Control signal procedures between AMF, SMF and
UPF are implemented, while less important parts are omitted such as security, policy, and so on. And
SDN/NFV module is an essential function for the 5G core networks. The green modules of SDN/NFV
and Virtualization delay blocks are implemented for new 5G features. So far, the SDN / NFV module
has been developed based on the existing 4G EPC network, but the 5G core entity will be developed
and applied through continuous updating.

3.5.1 Dual radio protocol architecture

Figure 4 User plane diagram of LTE-NR dual connectivity in downlink case

This model includes the LTE radio protocol stack and the 5G new radio (NR) protocol stack. These
entities reside entirely within the UE and the LTE eNB/NR eNB nodes. This model is based on dual

18
 connectivity, providing the cooperating network architecture between 4G LTE and 5G NR in the
transition period from 4G to 5G. Figure 5 shows the block diagram of LTE-NR dual connectivity for
user plane in downlink case. The LTE eNB is deployed as a C-plane anchor node (master eNB, shortly
MeNB) and the NR eNB is deployed for boosting the user throughput or balancing load between LTE
eNB and NR eNB (secondary eNB, shortly SeNB). The downlink traffic splits at the PDCP entity at
LTE eNB and routing to either the RLC entity at LTE eNB or that at NR eNB. In order to enjoy the
dual connectivity, the traffic split function is deployed at MeNB. The traffic split entity is defined for
all traffic, i.e., split bearer, individually. The PDCP TX entity, split layer, should perform the packet
sequencing for the PDCP RX entity to re-order the split packets. The PDCP RX entity, aggregation
layer, should perform the packet reordering to guarantee in-sequence deliver of the received packet to
upper layer. We have developed the dual radio protocol stack, the packet sequencing, packet reordering,
and simple traffic split algorithm exploiting the LTE model, developed by LENA, and the mmWave
radio model, developed by New York university (NYU). We also have developed the radio protocol
and the channel model for NR operating at the frequency below 6 GHz, presented in next chapter.
In this scenario, we can evaluate the performance of the protocol for LTE-NR dual connectivity.
For example, the following features can be evaluated: traffic split/routing algorithm, RLC queue
management scheme, MAC scheduling algorithm, and so on. These algorithms and schemes can be
modified or developed by user. The simulation results are as follows: end-to-end performance, such
as TCP/UDP throughput, round trip time, and congestion window size, specific protocol performance,
such as PDCP packet drop rate, RLC queue size/delay, PHY SINR values, and so forth.

3.5.2 NR radio protocol

We have developed the NR protocol stack that is compatible with the legacy LENA LTE module
and NYU mmWave radio model. We comprised the developed radio models above and the new radio
model operates at below 6 GHz that we implemented. This model calculates spectrum values like
power spectral density by applying different functions that calculate the propagation loss based on
specifications of 3GPP for each carrier frequency. In other words, the Channels between UEs and BSs
can be configured by selecting the appropriate scenarios according to the network environment to be
simulated. There are several scenarios classified at the frequency below 6 GHz such as urban micro
cell with high UE density, urban macro cell with high UE density, urban macro cell with one high-rise
per sector and 300m ISD and indoor hotspot cell with high indoor UE density. The scenarios can be
changed easily by the user of the simulator, resulting in different channels and different outcomes. In
the figure 1 diagram, NR channel model indicates the above procedure and there will be further
updates on the layer2 functions of LTE UE and BS.
We also developed the Full Dimension (FD)-Multiple Input Multiple Output (MIMO) technique.
This technique is applied to channel with below 6 GHz and mmWave (28 GHz, etc). The FD-MIMO
is based on hybrid beamforming consisting of digital precoding and analog beamforming. When using
the FD-MIMO, various antenna parameters, i.e., the number of vertical and horizontal antenna
elements, the number of vertical and horizontal TXRUs, and rotation of antenna panel, can be applied
to UEs and BS. We added the analog beamforming model, i.e., 1-dimension full connection model.
19
 For the hybrid beamforming, we added a parameter that configures the number of layers in the BS.
Users who want to see MU-MIMO performance should create a new scheduler that schedules each
user’s packet to the specific layer.

3.5.3 5G Core (and SDN/NFV)

A 5G System (5GS) including a 5G Core network (5GC) is expected to utilize Network Function
Virtualization (NFV) and Software Defined Networking (SDN) techniques. More specifically, in a 4G
System, the Evolved Packet Core (EPC) network functions are implemented on black boxes, however,
in the 5GS, network functions become softwarization, not implemented on black box hardwares. Each
network functions are installed and run on Virtual Machines (VMs) of NFV platforms, instead of black
boxes. And, these VMs can have connectivity with each other utilizing SDN technique, while
monitoring and controlling network traffic among them. These changes make network operators be
able to manage networks more easily and flexibly. For example, if a network function run on a VM
becomes overloaded due to too high computational workload, a network operator can run another VM
through an NFV platform. Then it can migrate some part of workloads from the original network
function to newly booted one. However, by the NFV/SDN techniques, additional delay caused by
provisioning and migrating VM is added to end-to-end communication delay. Our SDN/NFV
components include the function which measures these side effects of introducing virtualization and
SDN techniques to 5GC of 5GS.
Delay side effect means additional delays components compared to non-virtualized core networks.
Additional delays depending on where network operators put VNFs (VNF topology) can be evaluated.
For example, scaling delay can be caused by introducing NFV technique to 5G core network. With
NFV, each of 5G core entity could be run on VMs, forming VNF. If the VNF has 8 CPUs but its
workload exceed it, NFV platform auto-scales the VNF. In this case, the VNF is scaled-out, so two
VNFs might be operated. This auto-scaling takes some delays, so user might experience longer latency
due to NFV technique.
Operation side effect means that with SDN, we can control network traffic flowing through core
networks. Traffic re-routing or balancing operation can be evaluated. The simulation results are end-
to-end performances including VM delay, network delay, and so on. User can engineer traffic which
flow through SDN networks, and verify it with SDN switch throughput per port.

20
 Figure 5. SDN/NFV simulation process

The simulator reflects such side effects by following simulation process of figure 6. First, users
should set their own simulation topology and parameters related to SDN/NFV operation. Once they
(user codes) are provided to the simulator, it runs in two ways, virtualization modules and SDN
modules.
Virtualization modules calculate virtualization-related delays such as scaling delay or provisioning
delay. First, simulation nodes are placed depending on the topology of user codes. Dynamic workloads
(Simulation timely changes of workloads) of core NFs (MME, P/S-GW) are generated by using static
workloads of parameters of user codes. VNF delays are calculated after configurating scaling
thresholds and analyzing the topology, VNF policy.
SDN modules also places simulation nodes first. They configure OpenFlow switches and a
controller. After then they run OpenFlow application such as QoS bandwidth controller. The final
simulation results come out by merging results of virtualization modules and SDN modules.
The 5G architecture is defined as service-based and the interaction between network functions is
represented in two ways. In this paper, we will describe 5G core architecture with reference point
representation.

21
 NSSF AUSF N13 UDM

N22 N12 N8 N10

AMF N11 SMF N7 PCF N5 AF

N14 N15

N1 N2 N4

UE (R)AN N3 UPF N6 DN

N9

As above figure shows, the 5G System architecture consists of the following network functions
(NF).
Authentication Server Function (AUSF), Core Access and Mobility Management Function (AMF),
Data network (DN), Unstructured Data Storage network function (UDSF), Network Exposure
Function (NEF), NF Repository Function (NRF), Network Slice Selection Function (NSSF), Policy
Control function (PCF), Session Management Function (SMF), Unified Data Management (UDM),
Unified Data Repository (UDR), User plane Function (UPF), Application Function (AF), User
Equipment (UE), (Radio) Access Network ((R)AN), 5G-Equipment Identity Register (5G-EIR)
However, for simulation purpose, our simulator includes very specific and important parts of NFs,
UE, RAN, AMF, SMF and UPF. And only AMF, SMF and UPF are the NFs in 5G core. Role of each
NF is defined as follows.
The Access and Mobility Management function (AMF) includes the following functionality. Some
or all of the AMF functionalities may be supported in a single instance of an AMF:
- Registration management.
- Connection management.
- Mobility Management.
The Session Management function (SMF) includes the following functionality. Some or all of the
SMF functionalities may be supported in a single instance of a SMF:

22
 - Session Management e.g. Session establishment, modify and release, including tunnel
maintain between UPF and AN node.
- UE IP address allocation & management (incl optional Authorization).
- Selection and control of UP function.
- Configures traffic steering at UPF to route traffic to proper destination.
- Control and coordination of charging data collection at UPF.
- Downlink Data Notification.
The User plane function (UPF) includes the following functionality. Some or all of the UPF
functionalities may be supported in a single instance of a UPF:
- Anchor point for Intra-/Inter-RAT mobility (when applicable).
- External PDU session point of interconnect to Data Network.
- Packet routing & forwarding.
- Packet inspection and User plane part of Policy rule enforcement.
- Traffic usage reporting.
- Downlink packet buffering and downlink data notification triggering.

4. Parameter description
4.1 General parameters
 X2 interface settings

– X2 link data rate: configure data rate of X2 link, link between base stations (eNB, gNB).

 ex: 10Gb/s (There is no recommendation in 3GPP specification)

– X2 link delay: configure X2 link delay.

 ex: 30ms (There is no recommendation in 3GPP specification)

– MTU of X2 link: configure maximum transmission unit (byte) of X2 link.

 ex: 1500 (There is no recommendation in 3GPP specification)

 S1AP interface settings

– S1AP link data rate: configure data rate of S1AP link, link between MME and base stations

23
 (eNB, gNB).

 ex: 100Gb/s (There is no recommendation in 3GPP specification)

– S1AP link delay: configure delay of S1AP link.

 ex: 10ms (There is no recommendation in 3GPP specification)

– MTU of S1AP link: configure maximum transmission unit (byte) of S1AP link.

 ex: 1500 (There is no recommendation in 3GPP specification)

 S1-U interface settings

– S1-U link data rate: configure data rate of S1-U link, link between S-GW and base stations
(eNB, gNB).

 ex: 100Gb/s (There is no recommendation in 3GPP specification)

– S1-U link delay: configure delay of S1-U link.

 ex: 10ms (There is no recommendation in 3GPP specification)

– MTU of S1-U link: configure maximum transmission unit (byte) of S1-U link.

 ex: 1500 (There is no recommendation in 3GPP specification)

4.2 5G NR NSA
 Type of dual connectivity: select dual connectivity alternative, 1A or 3C

– Alternative 1A

– Alternative 3C

 PDCP reordering settings

– PDCP reordering enabler: PDCP sequence numbering/reordering functions enable or not.

 Enable: PDCP reordering function at RX entity is activated.

 Disable: PDCP reordering function at RX entity is inactivated.

– PDCP reordering timer: configure PDCP reordering timer.

 {0, 20, 40, 60, 80, 100, 120, 140, 160, 180, 200, 220, 240, 260, 280, 300, 500, 750} ms
24
  Type of splitting algorithms: for 3C, select way of traffic splitting.

– It is not 3GPP standardized techniques, it is just a traffic management technique on LTE-

NR dual connectivity.

– Splitting to single eNB: MeNB steers the traffic (PDCP PDU) to own RLC or SeNB’s RLC

– Alternative splitting scheme to two eNB: MeNB distributes PDCP PDU to own RLC or
SeNB’s RLC one by one.

– Assistant info-based splitting scheme: MeNB can calculate split ratio based on assistant
information, such as individual link performance and so on.

4.3 5G NR SA
 Operating band parameters

– FrequencyRange: Definition of frequency ranges. Currently, only FR1 is supported.

 [38.101-1] Table 1: Definition of frequency ranges

Frequency range Corresponding frequency range Parameter
designation Value
FR1 450 MHz – 6000 MHz 1
FR2 24250 MHz – 52600 MHz 2

– CenterFrequency: The center frequency in Hz. If there is no input by user, it is determined

by automatically calculating center frequency of operating band from following parameter
NrOperatingBand1.

– NrOperatingBand1: NR operating bands in FR1. For now, only TDD mode bands are
supported.

 [38.101-1] Table 2: NR operating bands in FR1

25
 NR Uplink (UL) operating band Downlink (DL) operating Duplex Parameter
Operating BS receive band Mode Value
Band UE transmit BS transmit
UE receive
FUL_low – FUL_high FDL_low – FDL_high

n38 2570 MHz – 2620 MHz 2570 MHz – 2620 MHz TDD 38
n41 2496 MHz – 2690 MHz 2496 MHz – 2690 MHz TDD 41
n50 1432 MHz – 1517 MHz 1432 MHz – 1517 MHz TDD 50
n51 1427 MHz – 1432 MHz 1427 MHz – 1432 MHz TDD 51
n78 3300 MHz – 3800 MHz 3300 MHz – 3800 MHz TDD 78
n77 3300 MHz – 4200 MHz 3300 MHz – 4200 MHz TDD 77
n79 4400 MHz – 5000 MHz 4400 MHz – 5000 MHz TDD 79

– NrOperatingBand2: NR operating bands in FR2. Not yet supported.

 Channel parameters

– PathlossModel: The type of path-loss model to be used.

– Scenario: The available channel scenarios are 'RMa', 'UMa', 'UMi-StreetCanyon', 'InH-
OfficeMixed', 'InH-OfficeOpen', 'InH-ShoppingMall'.

– ChannelCondition: 'l' for LOS, 'n' for NLOS, 'a' for all.

– OptionalNlos: Use the optional NLoS propagation loss model

 Frame structure and physical resources parameters

– NumerologyIndex: In this B6NR module, multiple types of SCS are available. In the table
below, µ indicates this parameter. The numerology index from 0 to 2 represent 15kHz,
30kHz, 60kHz SCS each. Only from 0 to 2 value of index are supported in FR1. As shown
in the table below, the number of subframes per frame is fixed at 10, and the number of
slots per subframe is determined by this value, so the number of slots per frame is also
assigned. Additionally, OFDM symbol duration and cyclic prefix duration values are
calculated by the input value of numerology index.

 [38.211] Table 3: Number of OFDM symbols per slot, slots per frame, and slots per
subframe for normal

26
 µ slot
N symb frame,µ
N slot subframe,µ
N slot
0 14 10 1
1 14 20 2
2 14 40 4
3 14 80 8
4 14 160 16
5 14 320 32

– ChannelBandwidth: In the FR1 band, the channel bandwidth has various options as shown
below. The number of RBs is determined by SCS and channel bandwidth. The channel
bandwidth for each NR band is limited as shown in the 2nd table below.

 [38.101-1] Table 4: Maximum transmission bandwidth configuration NRB

SCS 5MHz 10MHz 15MHz 30 MHz 20 MHz 25 MHz 40 MHz 50MHz 60 MHz 70 MHz 80 MHz 90 MHz 100 MHz
[kHz]
NRB NRB NRB NRB NRB NRB NRB NRB NRB NRB NRB NRB NRB
15 25 52 79 [160] 106 133 216 270 N.A N.A N.A N.A N.A
30 11 24 38 [78] 51 65 106 133 162 [189] 217 [245] 273
60 N.A 11 18 [38] 24 31 51 65 79 [93] 107 [121] 135

 [38.101-1] Table 5 Channel Bandwidths for Each NR band

NR band / SCS / BS channel bandwidth
NR SCS 5 10 15 20 25 30 40 50 60 70 80 90 100
Band kHz MHz MHz MHz MHz MHz MHz MHz MHz MHz MHz MHz MHz MHz
n38 15 Yes Yes Yes Yes
30 Yes Yes Yes
60 Yes Yes Yes
n41 15 Yes Yes Yes Yes Yes
30 Yes Yes Yes Yes Yes Yes Yes Yes
60 Yes Yes Yes Yes Yes Yes Yes Yes
n50 15 Yes Yes Yes Yes Yes Yes
30 Yes Yes Yes Yes Yes Yes Yes
60 Yes Yes Yes
n51 15 Yes
15 Yes Yes Yes Yes Yes
n77 30 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
60 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
15 Yes Yes Yes Yes Yes
n78 30 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
60 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
15 Yes Yes
n79 30 Yes Yes Yes Yes Yes
60 Yes Yes Yes Yes Yes

– SymbolsPerSlot: Currently, this parameter has a fixed value of 14. In the next version, it
27
 may include less OFDM symbols in the case of a mini-slot.

– SlotFormat: There are a lot of formats defined in 38.211. This would be similar to LTE-
TDD subframe configuration. In NR slot format, however, DL and UL assignment vary
along to the symbol level, not to subframe level and there are much more diverse formats
for various requirements. Currently, DL and UL symbols are supported and flexible symbol
needs to be further developed.

 [38.211] Table 6: Slot formats

Format Symbol number in a slot
0 1 2 3 4 5 6 7 8 9 10 11 12 13

0 D D D D D D D D D D D D D D
1 U U U U U U U U U U U U U U

 gNB-side parameters

– TxPower: Transmission power in dBm.

 UE-side parameters

TxPower: Transmission power in dBm.

4.4 5G Core (and SDN/NFV)

 AMF-related parameters

– AMF cpu capacity

 Decide computational power of virtual machine(s) running AMF(s).

– AMF memory capacity

 Decide memory size of virtual machine(s) running AMF(s). Memory capacity affects
migration delay or scaling delay, and so on.

– AMF disk capacity

 Decide hard disk size of virtual machine(s) running AMF(s).

– AMF bandwidth

 Decide link bandwidth of virtual machine(s) running AMF(s). Bandwidth affects almost
all virtualization delays such as scaling in/out delays.

– AMF workloads
28
  Decide expected amount of computational, memory, disk, and bandwidth loads which
will be taken to the AMF(s).

– AMF x axis

 Decide AMF’s x axis on simulation space.

– AMF y axis

 Decide AMF’s x axis on simulation space.

 SMF-related parameters (SMF-related parameters are similar to those of AMF)

– SMF cpu capacity

– SMF memory capacity

– SMF disk capacity

– SMF bandwidth

– SMF workloads

– SMF x axis

– SMF y axis

 UPF-related parameters (UPF-related parameters are similar to those of AMF)

– UPF cpu capacity

– UPF memory capacity

– UPF disk capacity

– UPF bandwidth

– UPF workloads

– UPF x axis

– UPF y axis

 Virtualization parameters

29
 – Scaling policy (memory split ratio)

 Decide when scaling operation should occur

 Decide amount of memory should be migrated to newly created VM.

– VM provisioning delay

 Decide the delay of provisioning new VM. This parameter can be replaced later by
introducing scheduling parameters.

 Data center parameters

– Data center topology

 Node ID

 Describe each node’s ID.

 Node ToR ID

 Describe ToR ID of each node. Each node is under ToR switches, and this parameter
indicates which ToR switch the node is connected.

 Node physical machine ID

 Describe physical machine ID of each node. In the data center, a physical machine can
operate several virtual machines. In the simulation, node is a VM. This parameter
indicates which physical machine the VM node is operated.

 Link bandwidth

 Bandwidth of each link.

 Link edge nodes

 Two nodes connected with the link.

 Software-Defined Networking (SDN) parameters

– Inter-switch data rate

 Link bandwidth between SDN switches.

– gNB-Switch-GW data rate

 Link bandwidth between SDN switch, gNB and GW nodes.

30
 – remote host-GW data rate

 Link bandwidth between UEs and GW nodes.

– In-port, out-port

 Decide a port which traffic enters in.

 Decide a port which traffic exit.

– Output port group

 Output port group for QoS control.

– QoS weight

 Give weight to each SDN forwarding rule for QoS control.

5. Test scenarios
5.1 Overview

Figure 6 Code block diagram

31
 Dual connectivity

– In the test scenario, we verify the setup procedure of LTE-NR dual connectivity, traffic
splitting in PDCP TX entity, and traffic merging in PDCP RX entity.

– To this end, we deploy a LTE eNB as master node, a NR gNB as secondary node, and UE
equipping the dual radio interface of LTE and NR.

– We consider the downlink case for testing the above.

– The PDCP TX entity splits the traffic based on split ratio calculated by using assistance
information from LTE and NR RLC entities, and routes split traffic onto RLC TX entities
in eNB and gNB. The PDCP TX entity allocates the sequence number to the spilt traffic in
order to the PDCP RX entity can merge and deliver the split traffic in sequence.

– The PDCP RX entity receives the data from LTE-NR dual inks and merges it in sequence
to deliver it to App layer.

– Based on LTE-NR dual connectivity, mobility algorithm can be evaluated.

 5G NR

– In the test scenario, we check out the SNR values of received transport blocks (TBs)
whether the building exists or not.

– A gNB is fixed on the y-axis, and its height is 21.5 m and a UE moves at a speed of 60
km/h in a direction parallel to the y-axis.

– Downlink packets for the UE are continuously generated during the simulation time, which
is set to 6 s.

– We set analog beamforming vector update period of which value is denoted by T, to 160
ms.

 5G Core (and SDN/NFV)

– In the test scenario, additional delays (scaling-in delay, scaling-out delay, migration delay,
etc.) from virtualization or traffic control using SDN are verified.

– We set the downlink test environment as follows.

 Two UEs are connected to an gNB.

32
  The gNB is connected to the SDN network consisting of a OpenFlow controller and
OpenFlow switches.

 SMF and UPF are connected to the SDN network.

 A remote host (server) is connected to the UPF.

– Virtualization effects are calculated by virt-5gc modules considering topology inputs and
VM inputs of user code.

– SDN traffic control is affected from qos-controller of OpenFlow modules.

5.2 Parameter setting

 Table 7: Simulation parameter settings
Parameter Value

LTE downlink bandwidth 20Mhz

LTE link capacity 75Mbps

LTE downlink carrier frequency 2110Mhz

mmWave bandwidth 1Ghz

mmWave carrier frequency 28Ghz

X2 DataRate 10Gb/s

X2 LinkDelay 20ms

TCP type TCP new reno

TCP segment size 1400byte

Transport layer (TCP/UDP) TCP send buffer size 30Mbyte

TCP receive buffer size 30Mbyte

UDP source rate 300Mbps

RLC mode RLC UM

mmWave scheduler mmWaveFlexTtiMaxWeightScheduler

LTE scheduler PfFfMacScheduler

RLC UM buffer size 100Mbyte

Source data rate 300Mbps

S1uLinkDelay 1ms

S1apLinkDelay 0s

33
Simulation time 20s

FrequencyRange 1

CenterFrequency 4700
Operating band parameters
NrOperationBand1 79

NrOperationBand2 n/a

PathlossModel Nr3gppPropagationLossModel

Scenario UMa
Channel parameters
ChannelCondition 1

OptionalNlos true

NumerologyIndex: 2

ChannelBandwidth 100
Frame structure and physical
resources parameters SymbolsPerSlot 14

SlotFormat 0

gNB-side parameter TxPower (dBm) 30

UE-side parameter TxPower (dBm) 30

UE speed (m/s) 1.5

Mobility parameter
H/O decision criteria SINR

Scaling policy (memory split ratio, %) 50

Virtualization parameters
VM provisioning delay (ms) 0

Inter-switch data rate 10Mbps

SDN parameters Switch-gNB data rate 100Mbps

Switch-GW data rate 100Mbps

UPF & SMF CPU [ea]

205/210
(Workload/Capacity)
Mem [MB]
910/916
(Workload/Capacity
Disk [GB]
Heavy load 3200/4000
(Workload/Capacity)
Bandwidth (Mbps)
9500/10000
(Workload/Capacity)
AMF CPU [ea]
195/200
(Workload/Capacity)

34
 Mem [MB]
505/512
(Workload/Capacity
Disk [GB]
1900/3000
(Workload/Capacity)
Bandwidth (Mbps)
9500/10000
(Workload/Capacity)
Light load UPF & SMF CPU [ea]
170/210
(Workload/Capacity)
Mem [MB]
860/916
(Workload/Capacity
Disk [GB]
2400/4000
(Workload/Capacity)
Bandwidth (Mbps)
9500/10000
(Workload/Capacity)
AMF CPU [ea]
160/200
(Workload/Capacity)
Mem [MB]
460/512
(Workload/Capacity
Disk [GB]
1300/3000
(Workload/Capacity)
Bandwidth (Mbps)
9500/10000
(Workload/Capacity)

5.3 Results
Figure 6 is the collection of graphs, which show the results of our NR non-standalone scenario simulation.
The figure 6 is plotted by gnuplot tool for depicting our simulation data.
Above figure consists of 4 graphs: the variation of RLC TX queue size, TCP RTT, RLC Path throughput
and TCP Cwnd. In all graphs, x axis represents simulation time.
As you know, because LTE link capacity is only 75Mbps, we cannot achieve more than 100Mbps
throughput. However, using 5G NR (mmWave) , we get more than 100Mbps throughput. You can find RLC
path throughput graph from above figures, and total throughput (LTE + mmWave) is 300Mbps that LTE
link cannot get. Therefore, by aggregating both bandwidths (LTE + 5G NR) in the dual connectivity
structure, we are able to increase UE’s throughput a lot.
One another benefit would be short latency due to UE’s connection to 2 eNBs. When downloading big file
or web page from server, dividing traffic into 2 eNBs would reduce overall latency a lot, compared with
using single eNB.

35
Moreover, in LTE-NR dual connectivity scenario, a mobility algorithm can be evaluated. Figure 8 shows
the results of mobility performances in LTE-NR dual connectivity. The left figure describes the RLC queue
backlog of each gNB in Byte with respect to time (sec). From this figure we can observe that gNB handover
is well performed with respect to UE mobility. The right figure shows the RLC throughput at the UE with
respect to time (sec). From this figure, we can observe that our NR handover procedure and algorithm is
suitable to support 5G UE.

Figure 7 Simulation results of LTE-NR dual connectivity

36
 Figure 8 Mobility performance in LTE-NR dual connectivity

Using the simulator for SDN/NFV module, user can get following graphs. In the scenario, an UE
communicates with a remote host, and core networks are virtualized. There are two conditions, heavy load
and light load. In heavy load condition, core network entities such as P/S-GW or MME are suffer from high
workloads. In light load condition, core network entities have lighter workloads than those of heavy load
condition.

Figure 9 User RTT in light load condition and heavy load condition

We can find out that with heavy load condition, user experienced more out-lied RTT compared to
light load condition. The out-lied points mean that auto-scaling is performed by NFV platform, so user
experience high RTT because packets are stalled during auto-scaling of VNFs.

37
 (a) Without building (b) With building

Figure 10 SNR values of received TBs

Figure 10(a) shows the SNR values of received transport blocks (TBs) when the building does not
exist. In an ideal case, where we know the channel matrix during the simulation, the highest SNR values
can be obtained by the best analog bemaforming vectors. We set the analog beamforming vector update
period T=160 ms to reflect the beam tracking procedure. We observe that the SNR values are lower than
those of the ideal case. This is because the gNB and the UE transmit and receive data using analog
beamforming vectors based on past channel information. Figure 10(b) represents the SNR values of
received TBs when the building exists. In both the ideal and non-ideal cases, we observe the SNR decreases
sharply when the channel condition is changed to the non-line-of-sight (NLOS) due to the building. When
the T is applied, if the channel condition is changed, i.e., from LOS to NLOS, vice versa, the SNR value is
zero. Therefore, we do not represent the minus infinity values on a log scale in Figure 10(b).

6. How to use
6.1 How to install 5G NR NSA simulator
Source code is in the server, 147.248.57.22:10022 at the folder, GigaKoreaProject
 Configure
 cd /GigaKoreaProject/mmwave_new_handover/ns3-mmwave/
 $ ./build.py
 $ cd ns-3.26
 ./waf configure

38
  Build
 $ ./waf –run scratch/mc-downlink
6.2 Workspace
Run program only in the scratch folder.
If you want to make your own simulation, you should move your code file to ~/ns3-mmwave/scratch/.
And when you run program, you need not write file extension (.cc)

Example
./waf --run scratch/example.cc
./waf --run example.cc
6.3 How to install 5G SDN/NFV simulator
5G SDN/NFV simulator is based on the ns-3 network simulator running on the linux operating system,
so we suppose that users who want to download source code and use it have experience on using linux.

1. Download source code from the 5G web site(http://5gopenplatform.org)

2. Extract source code on linux machine

$tar -xvzf [filename.zip] -C [extraction location]

3. Run configure file

$sh configure.sh
The configure script will automatically set below features.
- Essential packages for ns-3 will be installed
- Netbee and other packages for ofswitch13 will be installed
- ofswitch13 and ns-3 will be configured.
3.1. Download essential packages

39
$ sudo apt-get install build-essential gcc g++ python git mercurial unzip cmake
$ sudo apt-get install libpcap-dev libxerces-c-dev libpcre3-dev flex bison
$ sudo apt-get install pkg-config autoconf libtool libboost-dev

Type “Y” when the system asks you to install the packages

3.2. Configure SDN packages

3.2.1. Netbee library for ofswitch13
$ wget https://bitbucket.org/ljerezchaves/ofswitch13-module/downloads/nbeesrc.zip
$ unzip nbeesrc.zip
$ cd netbee/src/
$ cmake
$ make
$ sudo cp ../bin/libn*.so /usr/local/lib
$ sudo ldconfig
$ sudo cp -R ../include/* /usr/include/
3.2.2. ofswitch13 module
$ git clone https://github.com/ljerezchaves/ofsoftswitch13
$ cd ofsoftswitch13 $ git checkout v3.1.x
$ ./boot.sh
$ ./configure --enable-ns3-lib
$ make

40
 3.2.3. Link to ns-3
$ hg clone http://code.nsnam.org/ns-3.26
$ cd ns-3.26
$ hg clone https://bitbucket.org/ljerezchaves/ofswitch13-module src/ofswitch13
$ cd src/ofswitch13
$ hg update 3.1.0
$ cd ../../
$ patch -p1 < src/ofswitch13/utils/ofswitch13-src-3_26.patch
$ patch -p1 < src/ofswitch13/utils/ofswitch13-doc-3_26.patch

3.3. Configure the simulator

$ ./waf configure --with-ofswitch13=path/to/ofsoftswitch13
$ ./waf

6.4 How to run your own simulation

For running simulations for 5G SDN/NFV scenarios, users should make their own
simulation code in folder named “/scratch”.
For easy use, we prepare a basic simulation code for 5G SDN/NFV scenario in
“/scratch/virt-5gc-1to1-heavy2.cc”.

41
Users can modify the baseline scratch file for their customized simulation.
And if users just want to change the input parameter, “/input/virt-5gc-vmimput-
heavy.input” and “/input/virt-5gc-toposample.input” could be the keys.

42
The first one includes VM-related parameters such as cpu, memory, disk, bandwidth
capacities and workloads.

Figure 11 input/virt-5gc-vminput-heavy.input format

43
The second one includes topology parameters such as x, y axis of nodes, type of nodes,
and VM ID which node is placed on.

Figure 12 input/virt-5gc-toposample.input format

44
7. Change Log
 v1.1, 2019-02-17
 v1.2, 2019-04-15

45
8. References
[1] 3GPP, “Study on Architecture for Next Generation System,” 3rd Generation
Partnership Project (3GPP), TR 23.799 V14.0.0, Dec. 2016.
[2] 3GPP, "System Architecture for the 5G system", 3rd Generation Partnership Project
(3GPP), TS 23.501 V15.1.0, Mar. 2018.
[3] 3GPP, "Procedure for the 5G system", 3rd Generation Partnership Project (3GPP),
TS 23.502 V15.1.0, Mar. 2018.
[4] 3GPP, “Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved
Universal Terrestrial Radio Access Network (E-UTRAN),” 3rd Generation
Partnership Project (3GPP), TS 36.300 V15.0.0, Dec. 2017.
[5] 3GPP, “Packet Data Convergence Protocol (PDCP) specification,” 3rd Generation
Partnership Project (3GPP), TS 36.323 V14.2.0, Mar. 2017.
[6] 3GPP, “Radio Resource Control (RRC),” 3rd Generation Partnership Project (3GPP),
TS 36.331 V15.0.1, Jan. 2018.
[7] 3GPP, “Study on channel model for frequencies from 0.5 to 100 GHz,” 3rd
Generation Partnership Project (3GPP), TR 38.901 V14.3.0, Dec. 2017.
[8] 3GPP, “Multi-connectivity,” 3rd Generation Partnership Project (3GPP), TS 37.340
V15.0.0, Dec. 2017.
[9] ns-3 [online]. Available: https://www.nsnam.org
[10] Polese, Michele, et al. "Improved handover through dual connectivity in 5G
mmWave mobile networks." IEEE Journal on Selected Areas in Communications
35.9 (2017): 2069-2084.
[11] Calheiros, Rodrigo N., et al. "CloudSim: a toolkit for modeling and
simulation of cloud computing environments and evaluation of resource provisioning
algorithms." Software: Practice and experience 41.1 (2011): 23-50.
[12] Chaves, Luciano Jerez, Islene Calciolari Garcia, and Edmundo Roberto
Mauro Madeira. "Ofswitch13: Enhancing ns-3 with openflow 1.3 support."
Proceedings of the Workshop on ns-3. ACM, 2016.

46