Challenges in Migrating to LTE

Macquarie University
Contents Abstract 2

Faculty of Science ITEC850 Network Management Assignment #2: B3G or 4G Networks (Group Project) Due: 24th May 2012 Written by: Barry Ho & Willis Kwok Lecturer: Mr Milton Baar

Introduction
Aims and Goals Challenges Horizontal Handover Vertical Handover QoS Techniques Advantages/Disadvantages Conclusions & Future Work References Appendix A Appendix B

2
3 4 4 7 8 10 11 12 13 14

Appendix C

15

Abstract
There have been tremendous growths in the use of mobile technologies in recent years. In the early days, only the rich and keen early adopters were able to own mobile phones, which were large in size, required a substantial source of power to operate and were limited in functionality and quite restrictive in terms of mobility. Today, over 5.6 billion subscribed devices are in active use, which represent approximately 80% of the world population and are rising. (Gartner) Improved reception power coupled with increased network coverage and penetration, global roaming capability, sharp quality, fitto-palm size with large screen and lightweight are the significants of today’s user terminals. Given such advances, the growth within the large increases of cellular use has been on mobile data. In 2011, the total mobile data service revenues were close to $315 billion (Gartner). LTE is a serious improvement in network architecture to handle this surge in demand and is embraced by all providers. This paper first describes the background of LTE and then outlines the important challenges. Two of the challenges are related to the handover technologies and techniques on providing the required QoS for services, they are discussed in finer details.

1. Introduction
We are entering the fourth generation in mobile communications. We went from an analogue First Generation 1G system to a highly successful Second Generation (2G) with GSM which standardised global communications. GSM used digital voice and offered some basic data connectivity. Third Generation 3G (IMT-2000 with various UMTS technologies) offered the path to fast Internet. On the way to the 4G deployment (IMT-2000 Advanced), we are now in B3G (Beyond 3G phase) with operators migrating to LTE (3GPP Long Term Evolution). Section 2 discusses the aims and goals of LTE. Section 3 discusses the challenges for migrating to LTE. Section 4 and section 5 discusses technologies relating to horizontal and vertical handovers. Section 6 covers RAC (Radio Admission Control), DRA (Dynamic Resource Allocation) and PS (Packet Scheduling) which are functions in RRM (Radio Resource Management). These are the functions in the LTE roaming nodes (eNodeB) that are QoS related. Section 7 outlines the

2

advantages and disadvantages of the technology and what limitations the operators might see in implementing LTE. Finally, the paper concludes with the thoughts from the writers and what future works they think might be in the pipeline to make the Long Term Evolution worthwhile.

2. Aims and Goals of LTE
When GSM (Global System for Mobile communications) unified the global communications, making roaming beyond borders and boundaries possible, a new market has emerged – an expansion of service to go beyond voice call, text message, low data rate service (GSM/GPRS/EDGE), towards a multiservice air interface with backward compatibility available at ultra-high speed and yet, at a service cost lower than what its predecessors were charging with less deliverables. LTE was seen as a continuation to GSM and UMTS (Universal Mobile Telecommunications System) within 3GPP (3rd Generation Partnership Project), a key aim of UMTS and GPRS/EDGE was to deliver a multiservice air interface and quite naturally this had become the core aim of LTE, but LTE was designed from the start under the assumption that all services would be packetswitched. ([10] p.3) To deliver a mobile broadband service indistinguishable by the users between using it and the fixed line services such as ADSL (Asymmetric Digital Subscriber Line) and FTTH (Fibre-To-The-Home). To achieve such a period of no less their investments for LTE Release 8 • • • • • • • • task, and to ensure competitiveness for a than 10 years for the operators to recoup of installing the core networks, the targets were finalised in June 2005. ([10] p.7)

Reduced delays, in terms of both connection establishment and transmission latency Increased user data rates Increased cell-edge bit rate, for uniformity of service provision Reduced cost per bit, implying improved spectral efficiency Greater flexibility of spectrum usage, in both new and pre-existing bands Simplified network architecture Seamless mobility, including different radio-access technologies Reasonable power consumption for the mobile terminal

3

3. Challenges for Migrating to LTE
The data aspect has been the dominant factor of the LTE movement despite that only with the ability of converging data, voice and video on an all IP core (Evolved Packet Core) can truly migrate the legacy networks to LTE. To be able to handle voice before VoIP implementation is the first challenge most operators will face. 3GPP TS 23.273 [1] introduces a CSFB (Circuit Switched Fall Back) solution to address this challenge. CSFB enables a LTE UE (User Equipment) to register to a legacy CS (circuit switched) network so the CS network knows where to locate the device. With CSFB, the UE radio interface will switch to the legacy mode when making or accepting a voice call. The second challenge is, once IMS/VoIP support is added and with the LTE coverage is still limited, how to maintain a voice call when a UE moves out of the LTE coverage area. 3GPP TS 23.216 [2] offered a SR-VCC (Single Radio Voice Call Continuity) solution to address this challenge. Even for Mobile Data, handover is a general challenge which can get very complicated. This will be covered in the Horizontal handover and Vertical Handover sections. To effectively rollout LTE and not to face the traumatic financial burdens the 3G networks had once presented to their operators is yet another challenge. Some operators may consider the use of femtocells and to adopt an ‘inside-out’ strategy. LTE data rates are delivered in an indoor environment (e.g. home, hotspot, office, etc.) while the macro coverage is handled by the operator’s UMTS/HSPA networks. [12] QoS is another challenge for any packet data network when carrying multimedia data. This will be discussed in the "RAC, DRA and PS in Radio Resource Management (RRM)" section.

4. Horizontal Handover
Horizontal handover happens when a device associated with one wireless access point is transferred to another wireless access point using the same technology. This usually happens when the device is moving from an area covered by one cell to another area covered by another cell. There are also other causes like balancing of channel allocations (when one cell is too congested and some devices can be moved to adjacent cells) or when the handover is needed to avoid some interference issues.

4

In LTE network, horizontal handover happens when a UE switches the connection from one eNodeB to another eNodeB. Section 5.5.1 in TS 23.401 [3] presents an X2-based and a S1-based handover procedure. X2 is the interface between eNodeBs. S1 is the interface between an eNodeB and the EPC (Evolved Packet Core). S1-based handover is used when there is no IP connectivity between the target eNodeB and the source eNodeB. 4.1 X2-based handover The following diagram shows an X2-based handover when a UE is in an "RRC CONNECTED" state. Information for the handover events can also be found in Section 10.1.2.1.1 in TS 36.300 [4], which lists the handover preparation phase, handover execution phase and handover completion phase.

It starts when the UE moves from eNB1 coverage area to eNB2 coverage area. The UE sends a radio measurement report to eNB1 indicating that signal quality from eNB2 is better than signal

5

quality from eNB1. Once eNB1 makes the Handover decision, eNB1 starts the handover preparation phase by send out a Handover Request to eNB2 (1). eNB2 makes a Admission Control decision and sends a reply to eNB1 (2). eNB1 sends an RRCConnectionReconfiguration message including mobilityControlInformation as a handover command to the UE (3). When the UE successfully synchronize with eNB2, it sends out an RRCConnectionReconfigurationComplete message to the eNB2 (4). eNB2 then sends a "Path Switch Request" to MME (5). MME in turn sends a "Modify Bearer Request" to the Serving Gateway (6). Steps (7)and (8) are self-explanatory. Finally, in step 9, eNB2 sends a "UE Context Release" to eNB1 and triggers the release of resource by eNB1. Section 5.5.1.1.2 and 5.5.1.1.2 in TS 23.401 [3] show the two cases of whether the handover also involves the change of Serving Gateways. In the case that there is no change of Serving Gateway (like the example described above), MME sends out a "Modify Bearer Request" which switches the UE's GTC tunnel from the source eNodeB to the target eNodeB. In the case when changing Serving Gateway is required, MME uses a "Create Session Request" to establish a new GTC tunnel between the UE with the target S-GW and uses a Delete Session Request to remove the existing GTC tunnel on the source S-GW. 4.2 S1-based handover S1-based handover is demonstrated on Figure 5.5.1.2.2-1 on TS 23.401 [3]. Comparing with X2-based handover, in additional to creating new GTC tunnel between the target eNodeB and the target S-GW and deleting the original tunnel, it also creates two "Indirect Data Forwarding Tunnels" so that IP connectivity is continued (return traffic from PDN GW to the original eNodeB gets redirected to the target eNodeB). These "Indirect Data Forwarding Tunnels" are then deleted at the end of the steps because they are not needed anymore. 4.3 GTP or PMIP on S5/S8 As shown in Figure 4.2.1-1 and Figure 4.2.2-1 in TS 23.401 [3], the connection from Serving Gateway to PDN Gateway is the S5/S8 interface. The interface is S5 when the Serving Gateway and the PDN Gateway belong to the same operator (non-roaming situation). It is the roaming interface S8 when they belong to different operators (roaming with Home routed traffic). S5 and S8 can be GTP or PMIP based. TS 23.401 covers the use of GTP on S5/S8. TS 23.402 [5] covers the user of PMIP on S5/S8.

6

5. Vertical Handover
Vertical Handover happens when a device transfers its connection from a network using one technology to another network using another technology (also known as heterogeneous technologies). With LTE, it also includes the handover to older networks such as GSM, UMTS and CDMA2000 and vice versa. Handing over to other cellular technologies is common in LTE because the LTE network will have a smaller coverage than other legacy networks in the beginning. Vertical Handover also includes handover to and from WLAN or even Ethernet. Consider the scenario that you are in a video conference using your home network and you can just unplug the laptop and jump in a taxi to go to the airport and still have the video conference uninterrupted. 5.1 Vertical Handover with legacy 3GPP networks TS 23.401 [3] section 5.5.2 covers Vertical Handover procedures for LTE interworking with legacy 3GPP networks including UTRAN and GERAN (GSM EDGE Radio AN). In these cases, the S-GW and MME have two new interfaces (S3 and S4) to legacy SGSN node. For the case of UTRAN, RNC (Radio Network Controller) also has S12 interface to S-GW and so S-GW behaves like a legacy GGSN node. GERAN BSC (Base Station Controller) connects to SGSN for both control and user data and so it does not need to connect to S-GW directly. 5.2 Vertical Handover with non-3GPP networks TS 23.402 [5] covers vertical Handover solutions for LTE interworking with non-3GPP networks. While on the handover cases with legacy 3GPP networks, GTP or PMIP based Network Based Mobility are used and mobility is completely transparent to UE. For handover with non-3GPP networks, there are more options. Network-based mobility using PMIPv6 (RFC 5213 [7]) or host-based mobility using MIPv4 (RFC 5944 [8]) or DSMIPv6 (RFC 5555 [9]) can be used when LTE interworks with non-3GPP networks. Section 4.1.3 in TS 23.402 [5] describes that a static configuration or a selection method (IP Mobility management Selection / IPMS) can be used. Static configuration defines that the UE is configured to work with the type of mobility mechanism provisioned on the network. A risk of this is that if the configuration is not correct (a mismatch), then network handover will not work. On the other hand, if IPMS is used, then a mismatch will not happen.

7

TS 23.402 [5] listed out the following protocol options for handover with trusted or un-trusted non-3GPP networks.

Host based with DSMIPv6 Host based with MIPv4 FA mode Network based with PMIPv6 Network based with GTP or PMIPv6

Trusted Yes Yes Yes

Un-Trusted Yes Yes

More details about this can be found in appendix B.

6. RAC, DRA and PS in Radio Resource Management (RRM)
Section 16 in TS 36.300 [4] covered "Radio Resource Management (RRM) Aspects" for LTE. The following two RRM functions are QoS related and are handled in the roaming nodes (eNodeB). Radio Admission Control (RAC) Dynamic Resource Allocation (DRA) / Packet Scheduling (PS) 6.1 Bearer QoS parameters Section 13.2 in TS 36.300 [4] and section 4.7.3 in TS 23.401 [3] have good description about bearer QoS parameters. QoS parameters are used in Radio Admission Control and in packet scheduling decisions. QCI is one of the key QoS parameters and it is defined in TS 23.203 [6]. Section 6.1.7.2 in the specification listed that QCI includes the following. Resource Type (GBR or Non-GBR) Priority Packet Delay Budget Packet Error Loss Rate

For GBR bearer type, the following two parameters are associated with the bearer. - Guaranteed Bit Rate (GBR) - Maximum Bit Rate (MBR) 6.2 Radio Admission Control (RAC) The Layer 3 control protocol in the wireless link between UE and eNodeB is Radio Resource Control (RRC). When a UE sends out an

8

RRCConnectionRequest RRC command to a eNodeB, the roaming node needs to use some admission control algorithms to determine whether to send out an RRCConnectionSetup or an RRCConnectionReject reply. Section 16.1.2 in TS 36.300 [4] describes that, with a goal of providing high radio resource utilization (which means accepting new session whenever possible), RAC uses the following factors to make the decision. 1. the overall resource situation in the network (E-UTRAN). 2. the QoS requirements, the priority levels and the provided QoS in-progress sessions. 3. the QoS requirement of the new radio bearer request. However, the 3GPP documents do not list out the exact rules and algorithms for these decisions. There are many research papers on this and the implementations are vendor specific. 6.3 Dynamic Resource Allocation (DRA) / Packet Scheduling (PS) Radio (E-UTRAN) resources include Physical Resource Block (PRB) and Modulation Coding Scheme (MCS). Allocations of PRBs (and associated MCS) are in 1ms periods in the time domain. This is also known as TTI (transmission time interval). Appendix C contains more information about these. To schedule every PRB in a TTI, signalling in the eNodeB to UE direction using the PDCCH channel is required. For downlink scheduling, it tells a UE that the specific PRB is destined to it. For uplink scheduling, it tells a UE that it can use the specific PRB in a future TTI. It can be seen that for continuous transmission data like voice (voice over IP) where there is a frame every 10 to 20ms, a signalling flow will happen for every frame if it is handled in above way. Semi-persistent scheduling is used instead to solve this problem. According to information from TS 36.331 [13], a SPS-Config RRC setup command can be sent to MAC layer to establish a semi-persistent scheduling for a UE in either uplink or downlink direction, say, to use a PRB for one TTL in every four TTLs. This way, only one command signal flow in the PDCCH channel will be needed and the allocation will be persistent until a SPS-Config release command or until a certain time if implicitReleaseAfter is used in the setup command. This is a very effective QoS control for this type of traffic. 6.4 Scheduling decisions For downlink scheduling and link adaptation decisions, eNodeB makes use of "Channel Feedback Report" which comes from

9

information in CQI (Channel Quality Indicator), Rank Indicator (RI) and Pre-coding Matrix Indicator (PMI). RI and PMI are relevant to MIMO operation only. CQI and PMI can be of Wideband type (for entire system bandwidth) or of Multi-band type (for some frequency bands only). RI always describes the rank on the whole system band. For uplink scheduling and link adaptation decisions, SRS (Sounding Reference Signals) information is used. SRS is a reference signal transmitted on the last SC-FDMA symbol of every 1ms sub-frame on UE uplink transmission. Schedulers also need to allocate PRBs for HARQ (Hybrid Adaptive Repeat and Request) retransmissions. Various scheduling algorithms can be used in LTE systems including those based on proportional fair scheduler and round robin scheduler. Like Radio Admission Control, the 3GPP documents do not list out the exact rules and algorithms for these decisions.

7. Advantages, Disadvantages and Limitations
LTE is the answer to the exponential growth of data demand. If the operators cannot achieve a lower per bit cost with such a growth in demand by moving to LTE, ‘decoupling’ will occur between the cost of maintaining and upgrading the existing networks and the incremental revenues, resulting in poorer results on the operator’s bottom line. With the increased data rate, operators can introduce new offerings to generate new revenue stream. Because LTE is able to work in both TDD and FDD modes, it allows operators running different 2G and 3G technologies such as GSM, CDMA2000 and WCDMA to take an approach to LTE that suits their priority and capex restraints, especially when a lot of them are still recouping on their 3G investments. For those that have yet adopted 3G, the progression to LTE from a 2G legacy network is also a clear path to do catch-up even though the capital expenditure in this category is higher than a migration from HSPA+ network. However if the HSPA operators have limited spectrum they may be less inclined to migrate because the benefits of deploying LTE over a narrow band are not as significant. After a costly 3G lesson, operators could be more open in sharing network and spectrum to shorten the ROI period. The possibility of re-farming the 2G spectrums (e.g. 900MHz and 1800MHz) for use on LTE is also an advantage, however in doing

10

so may present problems when it comes to backward compatibility and whether it is in sync with other global operators. Refarming could also delay a LTE rollout as operators wait for 2G customers to migrate so they can phase out legacy networks that are still running on those bands. Decision by operators on which frequency bands to deploy LTE also has significant influences on equipment vendor in designing the LTE terminals. While operators are expected to rest with 900MHz, 1800MHz, 2.1GHz and 2.6GHz standards and that equipment vendors will have terminals compatible to these frequency bands, operators choose outside of these bands may find only limited number of terminals available and not be able to support roaming requests from guest users of other countries.

8. Conclusions and future work
The scene is set, the technology is ready, and the public has been notified and expectations are high, it’s decision time for the operators. A number of factors will affect the timing of the rollouts, most are commercially driven although the methods of migration can differ from carriers to carriers, most will take the approach that will give them the best competitive edge, best way to utilise the investments already made, and most importantly, not to lose market share let alone the intention to gain on its competitors. Making the right decisions is the key to success. Decisions on call admission control algorithms; decisions on the use of protocols where applicable (GTC or PMIP, CMIP or PMIP); decisions on the maximum number of resource blocks on uplink and downlink and what QoS features to offer; decisions on what type of handover to support and the architecture. Some of these decisions are straight forward because of the current constraints, but decisions made with prior constraints could prove to be bad ones in the future. Therefore forward thinking is needed in making such decisions. These include visions into the type of applications that may emerge in the future and the emerging technologies which might provide an even ‘longer term evolution’. Activities in LTE-Advanced and IMT-Advanced can help on these decisions. All of those that have been discussed in this paper are matter to and will be considered by the operators, their feedback will further influence the works at 3GPP as well as the equipment vendors, who are also in a battle within their rank to make the best out of the move to LTE.

11

The winners are the users, especially those who have taken the initiative to understand what LTE can give them, went ahead and selected the equipment and operator that provided the best offering to their needs.

References
[1] 3GPP TS 23.272 v10.7.0, 'Circuit Switched (CS) fallback in Evolved Packet System (EPS); Stage 2'. [2] 3GPP TS 23.216 v10.3.0, 'Single Radio Voice Call Continuity (SRVCC); Stage 2'. [3] 3GPP TS 23.401 v10.7.0, 'General Packet Radio Service (GPRS) enhancements for Evolved Universal Terrestrial Radio Access Network (E-UTRAN) access' [4] 3GPP TS 36.300 v10.7.0, 'Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Overall description; Stage 2' [5] 3GPP TS 23.402 v10.7.0, 'Architecture enhancements for non-3GPP accesses' [6] 3GPP TS 23.203 v10.6.0, 'Policy and charging control architecture' [7] IETF RFC 5213, 'Proxy Mobile IPv6' [8] IETF RFC 5944, 'IP Mobility Support for IPv4, revised' [9] IETF RFC 5555, 'Mobile IPv6 support for dual stack Hosts and Routers (DSMIPv6)' [10] LTE – The UMTS Long Term Evolution: From Theory to Practice by Stefania Sesia [11] LTE for UMTS OFDMA and SC-FDMA Based Radio Access by Harri Holma and Antti Toskala [12] LTE – Top 12 Challenges by Manish Singh http://www.wirelessweek.com/Articles/2009/09/LTE-Top-12Challenges/ [13] 3GPP TS 36.331 v10.5.0, 'Evolved Universal Terrestrial Radio Access (E-UTRA); Radio Resource Control (RRC); Protocol specification' [14] 3GPP TS 24.304 v10.0.0, 'Mobility management based on Mobile IPv4; User Equipment (UE) - foreign agent interface; Stage 3'

12

Appendix A
Project meeting minutes Week 5, scope of works was determined, a series of library books was selected and each member was given the task to focus on the four main topics that worth a total of 14 points. Week 6-7, readings, research works and phone discussions Week 8-9, write-ups were exchanged electronically and commented on. Week 10, proof reading began, abstract and introduction were added Week 11, paper finalised, further proof reading and format dress-up ready for submission Individual contributions Abstract – WK and BH Introduction – WK and BH Aims/goals – WK Challenges – WK and BH Horizontal and vertical handover – BH Techniques on QoS – BH Advantages/disadvantages – WK and BH Conclusions and future work – WK and BH

13

Appendix B – Mobility mechanism options for interworking with Trusted and Un-trusted non-3GPP networks
The following subsections list various mobility mechanisms options for interworking with Trusted and Un-Trusted Non-3GPP networks. 1. Host based mobility using DSMIPv6 for Trusted or Un-Trusted Non-3GPP In the architecture diagrams in Figure 4.2.2-2, 4.2.3-3 and 4.2.3-5 in TS 23.402, an S2c interface connects from UE to PDN Gateway. Section 6.1.2 and 7.1.2 in the document shows that this interface uses DSMIPv6. Section 6.1.2 is for Trusted Non-3GPP network and it is quite straight forward. Section 7.1.2 is for Un-Trusted Non-3GPP network and the protocol diagram as well as the previous architecture diagrams show that a new node ePDG (Evolved Packet Data Gateway) [section 4.3.4 in TS 23.402] is inserted between untrusted non-3GPPP network and PDN Gateway. As shown in Figure 7.1.2-1, IPSEC tunnel is used between UE and ePDG so that the traffic between them is secured over the untrusted network. 2. Host based mobility using MIPv4 FA mode for Trusted Non-3GPP network Host based mobility using MIPv4 FA mode is shown in Figure 6.1.1-2 on TS 23.402. This is one of the options for S2a interface between UE and PDN Gateway over a Trusted Non-3GPP network [Figure 4.2.2-1 on TS 23.402]. In this case, UE obtains an IP address from the Trusted Non-3GPP network, discovers the Foreign Agent (FA) and selects a care-of-address, and then registers it to the PDN Gateway which acts as the Home Agent. Full procedures are provided in section 5.1.2 on TS 24.304 [14]. 3. Network Based Mobility using PMIPv6 for Trusted Non-3GPP network Besides MIPv4 FA mode, for Trusted Non-3GPP network section 6.1.1 on TS 23.402 also lists that the S2a interface can also be using PMIPv6 [Figure 6.1.1-1 on TS 23.402]. In this case, the PDN GW function as the LMA (Local Mobility Anchor) in PMIPv6 [RFC 5213].

14

4. Network Based Mobility using GTP or PMIPv6 for Un-Trusted Non-3GPP For untrusted non-3GPPP network, the architecture diagram in Figure 4.2.2-1 on TS 23.402 shows that it uses a SWu interface from UE to ePDG over the untrusted non-3GPPP network and then with S2b interface from ePDG to PDN Gateway. Section 7.1.1 lists that the S2b can use either PMIPv6 or GTP for Network Based Mobility. The protocol diagrams in Figure 7.1.1-1 and 7.1.1-2 show that UE to ePDG path is encrypted using IKE2 and from ePDG to PDN Gateway, it can be GTP or PMIPv6.

Appendix C – PRB, MCS and TTI
Physical Resource Block (PRB) and Modulation Coding Scheme (MCS) are E-UTRAN Radio resources that can be made available to LTE user data communication. On downlink direction, each PRB includes 12 sub-carriers resulting 12 x 15 = 180KHz bandwidth. On uplink direction, each PRB also occupies 180 KHz bandwidth but it uses a single carrier method and not with sub-carriers. Obviously, because of the use of single carrier, multiple uplink PRBs from a UE will need to be in a continuous frequency spectrum. Downlink PRB groupings do not have this constraint. LTE specification defines that each device can have bandwidth between 1.4MHz and 20MHz. This is equivalent to between 6 and 100 PRBs. MCS refers to modulation (QPSK, 16QAM or 64QAM) and coding for the signal in the PRB. For downlink when there are sub-carriers, while it is possible that each sub-carrier has its own MCS but that would require too much feedback information and so MCS is applied to PRB level and not in sub-carrier level. MCS determines the bit rate of PRBs. Dynamic changes in MCS are also known as link adaptation. Allocations of resource blocks (and associated MCS) are in 1ms periods in the time domain. This is also known as TTI (transmission time interval). In the 3GPP specification descriptions, every 0.5 ms is a slot and every two slots form a transport block (with MAC header in front and padding at end). Transmissions in 10 ms (20 slots) forms one radio frame and the two slots (1ms) transport block is also known as a subframe.

15