Beruflich Dokumente
Kultur Dokumente
TABLE OF CONTENTS
Executive Summary ...................................................................................................................................... 2 1. Introduction .............................................................................................................................................. 3 1.1 Scope .................................................................................................................................................. 3 1.2 HSPA+ and LTE network deployment projections .............................................................................. 3 2. Multicarrier and multi-radio network evolution ......................................................................................... 6 2.1 Spectrum and deployment aspects ..................................................................................................... 6 2.2 HSPA evolution from multiple carriers to multicarrier .......................................................................... 8 2.3 LTE Evolution from multiple carriers to carrier aggregation .............................................................. 13 2.4 HSPA and LTE interworking.............................................................................................................. 16 2.5 HSPA+LTE carrier aggregation ......................................................................................................... 18 3. Benefits and use cases of HSPA+LTE aggregation .............................................................................. 19 3.1 Benefits of HSPA+LTE aggregation .................................................................................................. 19 3.2 Example USE cases for HSPA+LTE Aggregation ............................................................................ 20 4. HSPA+LTE aggregation system architecture considerations ................................................................ 21 4.1 Service or core network level split/merger ........................................................................................ 23 4.2 HSPA RAN level split/merger ............................................................................................................ 25 4.3 LTE RAN level split/merger ............................................................................................................... 28 5. Practical implementation aspects of HSPA+LTE aggregation............................................................... 30 5.1 Base station Radio implementation aspects ..................................................................................... 30 5.2 Device Radio implementation aspects .............................................................................................. 31 5.3 Implementation aspects other than radio processing........................................................................ 33 6. Conclusion ............................................................................................................................................. 34 Abbreviations .............................................................................................................................................. 35 Acknowledgements ..................................................................................................................................... 37
EXECUTIVE SUMMARY
LTE networks are being rolled out at an increasing speed, while at the same time the existing HSPA networks are expanded and upgraded with the more advanced HSPA+ features in order to cater to the ever-increasing appetite for wireless data. Due to the major investments in the HSPA+ infrastructure and the vast and rapidly increasing HSPA+ based mobile broadband device penetration the two networks can be foreseen to coexist in parallel for years to come. The evolution of both HSPA+ and LTE standards has introduced aggregation of carriers for higher data rates, better load balancing and increased spectrum utilization, and since the dawn of LTE, the standard support for radio level interworking for HSPA and LTE radios has been included. A natural continuation of such development is to tighten the interworking even further and introduce similar aggregation of carriers between the two radio access technologies. The HSPA+LTE aggregation allows for transmitting data to one user simultaneously using both the HSPA and the LTE radios for maximal utilization of the available spectrum and the deployed equipment. This is considered beneficial especially in the environment where the spectrum that needs to be shared between the two radio access technologies is not abundant, and the deployed HSPA and LTE capacities and user data rates suffer from spectrum crunch. One example of such deployment is the 900 MHz for HSPA and the 800 MHz for LTE which are both seen attractive bands for building the coverage due to the low frequency but also suffer from very limited spectrum availability. With aggregation of the two bands it is possible to provide the high data rates expected from the LTE services while at the same time maintain coverage for the HSPA devices. The same gain mechanisms that have been seen beneficial for Multicarrier HSDPA as well as LTE Carrier Aggregation can be benefited from by aggregating HSPA with LTE. At low or medium load, HSPA+LTE aggregation is able to take advantage of the unused resources leading to significant data rate increases both at the cell edge and the cell center for the carrier aggregation capable devices. In addition, the carrier aggregation enables fast (millisecond level) load balancing across the carriers thus improving the data rates of all users. A number of possible network architectures can be foreseen for HSPA+LTE aggregation, and are briefly touched upon in this white paper. Most promising architecture options are seen with co-located multiradio base stations with the base station (NodeB + eNodeB) acting as the data aggregation point, and simultaneously maintaining the existing network architecture for the devices connecting to the network with one radio system at a time only. This architecture can utilize some of the already deployed RF hardware in the base station, while new baseband functionality managing the data flow is required. On the device side, receiver radio architectures capable of multiband carrier aggregation should be suitable also for HSPA+LTE aggregation. While Dual-Cell HSDPA is already in commercial operation, and higher levels of HSPA carrier aggregation as well as LTE carrier aggregation are part of 3GPP specifications existing today, HSPA+LTE aggregation is currently not standardized. Although conceptually straightforward and building on already standardized concepts, HSPA+LTE aggregation is a major feature, with a standardization effort comparable to that of LTE carrier aggregation.
1.2 HSPA+ AND LTE NETWORK DEPLOYMENT PROJECTIONS HSPA, HSPA+ AND LTE DEPLOYMENTS AS OF JUNE 2012
HSPA: 473 commercial networks in 180 countries HSPA+: 227 commercial networks in 109 countries LTE: 91 commercial networks in 47 countries LTE: 335 operator Commitments worldwide LTE: Over 130 commercial networks expected by year end 2012
ubiquitous in major cities throughout Latin America and is setting the stage for future deployments of LTE in 2012 and beyond as operators gain access to new spectrum assets.
Clearwire is planning to launch LTE in TDD spectrum in 31 cities in the first half of 2013.
ABI Research predicts LTE mobile broadband lines will hit 80m by 2013. ABI Research, 24 October, 2011.
According to research from In-Stat, LTE mobile broadband technology is set for a surge in growth over the next four years. Between now and 2015, the number of people signed up for the next-generation service will increase by 3,400 percent, the market intelligence firm claimed.2 In-Stat attributed this rapid rise in subscriptions to consumer desire to connect to the Internet while on the move at any time of day and using a variety of devices, such as smartphones and tablet PCs. More than half of all infrastructure rollouts from network operators are now based on LTE, the organization revealed, sparking a decline in 2G usage from 2012 onwards.3 As LTE is gaining traction throughout the industry, it is also getting an increasingly larger chunk of network operator budgets. A recent report from IHS iSuppli indicates that spending on LTE infrastructure worldwide is set to more than triple from $8.7 billion in 2012 to $24.3 billion in 20134. IHS iSuppli further stated that there will be about 200 LTE networks operating commercially or being deployed around the world by next year, about 40 more than were in place in 2010. Research from In-Stat claims that tablets will have the highest 3G/4G attach rate among all cellularenabled portable and computing devices with 78 percent of tablets shipping with a 3G/4G modem by 2015. The research firm suggests that this trend represents an opportunity for mobile operators to move beyond the maturing handset market and into connecting emerging wireless device markets, like ereaders and tablets. A senior analyst at In-Stat predicts that by 2015, 65 percent of e-readers worldwide will ship with an embedded 3G/4G modem.5 The research firm also notes that approximately 16 million portable and computing devices shipped with 3G/4G cellular connectivity in 2010 and that over 50 percent of all 3G/4G tablets in 2015 will have LTE WAN connectivity. Finally, 4G Americas research shows 473 HSPA operators, of which 227 have deployed HSPA+. As of June 2012, there were 91 commercial deployments of LTE in 47 countries, with 335 total operator commitments to the technology.
2 3
LTE mobile broadband set for 3,400% growth by 2015. (In-Stat, June 2011) LTE mobile broadband set for 3,400% growth by 2015. (In-Stat, June 2011) 4 Fresh Research forecasts spending surge (IHS iSuppli, February 2012) 5 78% of tablets shipped in 2015 will have 3G/4G modem By eGov Innovation Editors | May 23, 2011
2. MULTICARRIER AND MULTI-RADIO NETWORK EVOLUTION 2.1 SPECTRUM AND DEPLOYMENT ASPECTS
Mobile operators are being driven to pursue carrier aggregation techniques by both technology and operational realities. Ever rising traffic volumes are motivating service providers towards technologies that exploit spectrum resources in the most efficient and economical manner. Spectrum holdings located across several frequency bands, and the coexistence of deployments based on diverse access technologies such as HSPA and LTE over long periods of time also incentive the use of carrier aggregation techniques. Broadly speaking, carrier aggregation technologies provide benefits such as the following Maximize the total peak data rate and throughput performance by combining peak capacities and throughput performance available at different frequencies Provide a higher and more consistent quality of service to customers as a result of load-balancing across frequencies and systems. A customer encountering congestion in one band and one system can seamlessly access unused capacity available at another frequency or system Mitigate the relative inefficiencies that may be inherent in wireless deployments in non-contiguous or narrow (5 MHz or less) channel bandwidths, often spread across different spectrum bands
The universe of potential frequencies that could potentially exploit carrier aggregation techniques is large. Most obviously, these include frequencies being used for IMT systems today. In the future, this should expand to include spectrum being contemplated for IMT-Advanced systems, as well as spectrum that may be re-farmed from GSM use toward more advanced technologies or other spectrum unlocked or re-farmed for WWAN usage. In the former category are spectrum bands common across many countries such as digital dividend spectrum (700 or 800 MHz depending on the ITU Region) and 2500 (also known as the 2600) MHz bands, as well as AWS (1700/2100 MHz) in the Americas (ITU Region 2). GSM spectrum that may be repurposed includes widely deployed bands such as the cellular and SMR bands (at 800-850 MHz) and 1900 MHz in the Americas, and 900 and 1800 MHz in other areas of the globe. Currently deployed spectrum bands differ widely in terms of contiguous bandwidth and in channelization schemes. Further, service providers hold much more paired than unpaired bandwidth. Compounding matters, bands allocated to mobile broadband are diversifying. All of these factors conspire to present growing challenges to equipment vendors and device OEMs developing multi-frequency (and increasingly multimode products). Mobile handheld devices present especially keen issues related to battery size limitations, screen size, weigh, and constrained interiors into which more and more RF components must be squeezed to accommodate increasing numbers of frequencies. . HSPA systems can only be deployed with carriers with nominal bandwidth of 5 MHz or multiples thereof (up to 40 MHz with 8 aggregated 5 MHz carriers) and only in paired mode, while LTE technology is specified for deployment in both unpaired and paired channels, and in a wide range of different channel widths from 1.4 MHz through 3, 5, 10 and 15 MHz options up to the maximum channel width of 20 MHz today, to 40 and even 100 MHz (as contemplated in LTE-Advanced with up to 5 aggregated 20 MHz carriers).
The efficiency of LTE is greater in the wider channel widths (2 x 10 MHz and greater). And 5 MHz HSPA channels combined with existing HSPA carrier aggregation and other technological advances can deliver performance that is competitive with todays LTE systems. HSPA systems will continue to be maintained for many years, as momentum behind LTE technology continually ramps up. Consequently, service providers will increasing be operating mixes of HSPA and LTE networks, across multiple bands. This reality suggests that operators and vendors should seriously consider the potential gains that may be had by combining the performance of existing systems via techniques such as HSPA+LTE carrier aggregation techniques. There is some commonality of frequencies within ITU regions, and to general, commonality exists to greater extent between ITU Regions regions and Region 2 (Americas). One exception is the 2500/2600 achieve widespread global use with the increasing adoption of the spectrum and a mid-band of 50 MHz unpaired spectrum). some extent between regions. In 1 and 3 than between these two MHz band, which is on a path to ITU Option 1 (2x70 MHz paired
In summary, there are a number of factors heightening the importance of carrier aggregation developments, including the potential benefits of HSPA/LTE carrier aggregation. These include Overlapping deployments of HSPA and LTE that will persist through the end of the decade, if not beyond. The need to maintain and enhance existing networks, both for service continuity to the installed base of device as well as to maximize returns on investment Varying technology features (bandwidth flexibility or limitations, channelization scheme, duplex options) Spectrum scheduled to be auctioned, as well as additional spectrum being pursued globally (i.e., post WRC-12) or regionally, which should be deployed in the most optimal way given existing network investments, capabilities, and limitations.
Rel7
HSPA+
Rel8
DCHSDPA
DBDCHSDPA
Rel9
Rel10
4CHSDPA
Rel11
8CHSDPA
Rel12
DCHSUPA
The concept of multiple carriers operation for HSPA was first introduced in Rel-8 as Dual-Cell HSDPA, with the scope of increasing coverage for high data rates in deployments where multiple carriers are available. DC-HSDPA operation is applied to two adjacent 5 MHz carriers and by scheduling HSDPA transmissions on both carriers simultaneously, allows doubling the peak data rate from a single HSPA+ carriers 21 Mbps to 42 Mbps with 64QAM when MIMO is not used. DC-HSDPA users can be scheduled on either of the two carriers, and either carrier can be configured as the primary serving cell, thereby benefiting an efficient load balancing between carriers. The two HS-DSCH transport blocks are processed independently, including the HARQ retransmissions. Rel-9 further extended the DC-HSDPA operation to be possible simultaneously with MIMO. 3GPP specifications define DC-HSDPA requirements for all the same frequency bands that have been defined for single carrier operation. Combining multiple carriers to Multicarrier HSDPA for a UE is performed only on the MAC-hs in the Node B, and there is a single RLC and PDCP layer just as with the single carrier operation, and practically the only difference in the RNC user plane when comparing to single carrier HSDPA is higher user throughput. At the MAC-hs layer in the Node B, each aggregated carrier has its own independent Hybrid Automatic Repeat reQuest (HARQ) entity. From a UE perspective, characteristics of each carrier procedures are unchanged with respect to basic single carrier HSDPA operation. Figure 2 shows the multicarrier mapping on downlink.
To allow multiple carriers operation in deployment scenarios when adjacent bands are not available, Dual-Band DC-HSDPA was introduced in Rel-9. The primary and secondary serving carriers reside in different bands, and the uplink transmission can be configured in either one of the two bands. The introduction of DB DC-HSDPA can be regarded as taking the evolution from multiple single carrier cell systems to multicarrier systems to include also the aggregation of non-contiguous spectrum bands. The capability of scheduling transmissions over multiple carriers of different bands provides an efficient utilization of the spectrum resources resulting in a substantial increase in cell capacity. In Rel-9 only three band combinations were allowed, and other band combinations have been added at a later stage while retaining the same functionalities as in Rel-9 specifications and can be implemented in Rel-9 networks and devices in a release independent manner. Currently defined band-combinations DB DC-HSDPA are listed in Table 1.
Table 1: 3GPP-defined Dual-Band Dual-Cell HSDPA band combinations
Band A I (2100 MHz) II (1900 MHz) I (2100 MHz) I (2100 MHz) II (1900 MHz)
Band B VIII (900 MHz) IV (1.7/2.1 GHz) V (850 MHz) XI (1500 MHz) V (850 MHz)
Figure adapted from 3GPP TS36.300 Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Overall description, V10.7.0 9 4G Americas HSPA+LTE Carrier Aggregation June 2012
Similar to DC-HSDPA, multiple carriers operation was introduced in the uplink in Rel-9 as DC-HSUPA, and the peak data rate doubled to 23 Mbps with 16QAM. DC-HSUPA users transmit two E-DCH transport blocks, one on each uplink carrier, and each transmission is done independently according to the principles used for the non-serving cells. The two carriers belong to the same sector of a serving NodeB, and the serving NodeB can activate/deactivate the secondary carrier dynamically. DC-HSUPA can only be used with DC-HSDPA because control signaling for the secondary UL carrier is carried over the secondary DL carrier. DC-HSDPA can instead be activated regardless if the uplink uses single or dual carrier(s). 3GPP specifications define DC-HSUPA requirements for all the same frequency bands that have been defined for single carrier operation.
8x5MHz
40MHz
8CHSDPA Rel11
4x5MHz
20MHz
4CHSDPA,Rel10 Noncontig.singleband,Rel11
2x5MHz
10MHz
DualCellHSDPA,Rel8 DualBand,Rel9
Figure 3: Aggregating more and more carriers to increase the total transmit bandwidth
Driven by an increasing demand for high data rates, multicarrier operation in the DL has evolved with the introduction of 4 carriers and 8 carriers in Rel-10 and Rel-11, respectively. The additional flexibility provided by the larger number of carriers improves the load balancing through the dynamic configuration of the serving cell of each multicarrier user. As is the case with DC-HSDPA, also with 4C-HSDPA and 8CHSDPA, all secondary carriers can be dynamically activated/deactivated by the serving NodeB through HS-SCCH orders. Depending on the type of traffic, the deactivation of all carriers in a frequency band can be useful for UE power savings. With the 4C-HSDPA feature, four HSDPA transmissions can be scheduled simultaneously over four carriers that do not need to be adjacent and can reside on different bands, featuring a peak data rate of 168 Mbps when configured with 2x2 MIMO and 64QAM. Similar to DC-HSDPA, each transmission is done independently and all secondary serving carriers can be activated/deactivated in a dynamic fashion by the serving NodeB. The uplink signaling, as in DC-HSDPA, is carried over a single carrier, and the feedback channel has been redesigned to include the information for all four DL transmissions. The band combinations for 4C-HSDPA include up to two frequency bands, and up to three carriers can be scheduled in the same band. All supported band combinations up to Release 10 require configuring adjacent carriers within each aggregated band to facilitate the UE receiver implementation. As for DB DCHSDPA, other band combinations can be added at a later stage. Currently defined band-combinations for 4C-HSDPA where carriers on a band are adjacent to each other are listed in Table 2.
Table 2: 3GPP-defined 4-Carrier HSDPA band combinations with all carriers within a band adjacent to each other
4C-HSDPA Configuration I-3 II-3 II-4 I-2 VIII-1 I-3 VIII-1 I-2 VIII-2 I-1 V-2 I-2 V-1 I-2 V-2 II-1 IV-2 II-2 IV-1 II-2 IV-2 II-1 V-2
I (2100 MHz)
V (850 MHz)
II (1900 MHz)
IV (1.7/2.1)
II (1900 MHz)
V (850 MHz)
Carrier combination 3 3 4 2+1 3+1 2+2 1+2 2+1 2+2 1+2 2+1 2+2 1+2
3GPP release Rel-10 Rel-11 Rel-11 Rel-10 Rel-10 Rel-11 Rel-10 Rel-10 Rel-11 Rel-10 Rel-10 Rel-10 Rel-11
3GPP Rel-11 further extended the supported cases for 4C-HSDPA to include single-band non-adjacent carrier configurations. In these cases all carriers of a 4C-HSDPA configuration reside in the same frequency band, but in two non-adjacent blocks. The carriers within each block are adjacent to each other, but there is a gap between the two blocks. Currently defined band-block combinations for the noncontiguous single-band 4C-HSDPA are listed in Table 3.
Table 3: 3GPP-defined 4-Carrier HSDPA single band non-adjacent carrier combinations
Single-band nonadjacent 4C-HSDPA Configuration I 1-5-1 I 1-5-2 I 1-10-3 IV 1-5-1 IV 1-10-2 IV 2-15-2 IV 2-20-1 IV 2-25-2
Band
Carrier combination 1+1 1+2 1+3 1+1 1+2 2+2 2+1 2+2
Gap between band blocks 5 MHz 5 MHz 10 MHz 5 MHz 10 MHz 15 MHz 20 MHz 25 MHz
3GPP release Rel-11 Rel-11 Rel-11 Rel-11 Rel-11 Rel-11 Rel-11 Rel-11
I (2100 MHz)
IV (1.7/2.1 GHz)
The introduction of 8C-HSDPA is a further extension of the multicarrier operation with eight carriers. Similar to the four carrier feature, in 8C-HSDPA the transmissions are independent. The carriers do not need to be adjacent and can reside on different frequency bands. The activation/deactivation of the secondary carriers is done by the serving NodeB through physical layer signaling. The uplink signaling is carried over a single carrier. The first band combination for 8C-HSDPA to be introduced in 3GPP is 8 adjacent carriers on band I (2100 MHz).
Release11
336Mbps
Release10
168Mbps
40MHz,2x2MIMO 20MHz,4x4MIMO
84Mbps
10MHz 2x2MIMO
20MHz 2x2MIMO
Release7
28Mbps
42Mbps
10MHz NoMIMO
Carrier aggregation for LTE is performed on the MAC and PHY layers only, and there is a single RLC and PDCP layer for all aggregated component carriers. At the MAC layer, each component carrier has its own independent Hybrid Automatic Repeat reQuest (HARQ) entity and physical layer. From a UE perspective, characteristics of the HARQ procedures for each component carrier are unchanged with respect to Release-8/9. Figure 6 shows the CC mapping on DL.
From the higher layer perspective, each component carrier appears as a separate cell with its physical cell identifier. Therefore, it appears that a carrier aggregation UE is connected to multiple cells. Among the multiple cells the UE is connected to, one particular cell is denoted as primary serving cell, while other cells (up to four) are denoted as secondary serving cells. Primary serving cell plays a unique and essential role with respect to security, upper layer system information, and some lower layer functions, while secondary serving cells are configured to primarily provide additional resources for UE to transmit and receive data. Another difference between primary and secondary serving cell is that primary serving cell can only be changed via RRC (re)configuration, while secondary serving cells, once configured via RRC signaling, can be activated or deactivated by MAC signaling without additional RRC signaling. This feature enables very fast activation and deactivation of secondary serving cells. One salient feature of LTE carrier aggregation is cross-carrier assignment, where DL scheduling or UL grant information of one component carrier can be carried via the PDCCH of another component carrier. Specifically, a PDCCH on one component carrier can schedule data transmissions on another component carrier by including a 3-bit Carrier Indicator Field (CIF) in the grant message to indicate the target component carrier. This is especially useful when secondary serving cell cannot be used to convey control information reliably. As an example, Figure 7 illustrates the regular DL assignment without crosscarrier assignment, while Figure 8 shows DL assignment with cross-carrier assignment, where PDCCH of component carrier 2 is used to schedule not only component carrier 2, but also component carrier 1 and 3.
Figure adapted from 3GPP TS36.300 Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Overall description, V10.7.0 14 4G Americas HSPA+LTE Carrier Aggregation June 2012
While Release-10 air interface allows up to five component carriers, only limited inter-band carrier aggregation combinations are defined (Table 5), and only intra-band carrier aggregation with contiguous component carriers for limited bands are defined (Table 4) as of Release-10. More band combinations are being defined in Release-11 and beyond.
CA_1-5
FDD
3GPP TS36.101 Evolved Universal Terrestrial Radio Access (E-UTRA); User Equipment (UE) radio transmission and reception, V10.6.0 15 4G Americas HSPA+LTE Carrier Aggregation June 2012
HSPA
Idle
LTE
Idle
Reselect
Connected
URA_PCH
Connected Connected
CELL_PCH
Redirect
CELL_FACH
CELL_DCH
Handover
Figure 9: The procedures for moving UEs between HSPA and LTE.
When the UE is in connected mode in LTE, it can be moved to HSPA by means of an inter-RAT handover or inter-RAT redirection procedures. An inter-RAT handover leads to a very short interrupt in the communication: in the order of tens of milliseconds. This is achieved through reservation of resources in the target cell before the serving cell is released. Also, data is forwarded from the serving RAT to the target RAT. After the inter-RAT handover procedure, the UE ends up in the RRC state CELL_DCH in WCDMA and the UE immediately proceeds to update the routing area so that it can be reached in the new RAT. The corresponding procedure can be used to move a UE in CELL_DCH in WCDMA to LTE. After the inter-RAT redirection procedure the UE ends up in the RRC idle mode and after finding the target cell proceeds to register to the WCDMA RAT with Cell Update procedure. The inter-RAT redirection procedure leads to significantly longer outage than the inter-RAT HO procedure: no data transmission is possible until the routing area update has been performed. The outage is in the order of a few seconds. When an HSPA UE is in any of the RRC states CELL_FACH, CELL_PCH or URA_PCH, it performs normal cell reselection. Here, if the UE reselects an LTE cell, the UE enters Idle mode and makes an access to LTE. In CELL_FACH, it is also possible to use inter-RAT redirection procedure. If the UE finds a cell where it was directed, this procedure performs relatively well. However, when the UE fails to find a cell, it needs to establish a connection to another cell, and this may take some time. In 3GPP Release 11, improvements to this redirection procedure are being discussed that will minimize the interrupt in case the redirection is unsuccessful. The transitions are depicted in Figure 9. The Operation and Maintenance (O&M) and Self Optimizing Network (SON) is another important aspect of interworking of HSPA and LTE radios. The O&M/SON management principle is shown in Figure 10. The same framework used to operate joint HSPA and LTE network deployments is naturally applicable also for the HSPA+LTE aggregating network. Depending on which radio aggregation architecture is considered, different O&M/SON improvements benefiting from tighter radio integration could be foreseen, although this aspect of the HSPA+LTE aggregation is not considered further in this paper.
Figure 10: SON umbrella for joint LTE and HSPA deployment
Similarly, when considering HSPA+LTE aggregation, one can think of both downlink and uplink as well as co-site and inter-site aggregation of the two technologies, but for the same reasons leading to being downlink-centric and emphasizing downlink aggregation, the main focus of the HSPA+LTE aggregation, at least initially, can be expected to be on intra-site aggregation of downlink carriers.
LTE
LTE evolution
HSPA
Rel-5Rel-9
Rel-7Rel-11
and beyond
Figure 11: 3GPP standard evolution of LTE carrier aggregation, HSPA carrier aggregation and HSPA+LTE interworking
3. BENEFITS AND USE CASES OF HSPA+LTE AGGREGATION 3.1 BENEFITS OF HSPA+LTE AGGREGATION
As discussed in sections 1 and 2, at least the initial focus of the HSPA+LTE aggregation is seen to be on aggregating co-sited downlink carriers. Respectively, the discussion of the benefits in this section is considering co-sited downlink carriers. HSPA+LTE aggregation utilizes same mechanisms as the intra-RAT carrier aggregation schemes described in section 2 and is thus expected to bring similar data rate gains: Data rates of carrier aggregation UEs boosted by utilizing unused resources from overlapping cell(s) operating on different carrier(s) Data rates of all UEs improved by fast (TTI level) load balancing
Similar to the intra-RAT carrier aggregation, the gains are highest at low/medium load and they benefit both the cell edge and the cell center UEs. At high load with multiple active UEs per cell it is possible to perform load balancing handovers to balance the load and thus aggregation of carriers is less beneficial. However, statistics from todays mature HSPA networks have shown that due to the burstyness of the data traffic there is often only one active UE per cell with data in the RAN buffers (even though there might be several UEs connected to the cell). In such case load balancing handovers are not helpful and part of the resources remain unused. Also if the data bursts are very short, the load balancing handovers can be rather inefficient due to the handover delays and overheads. In such scenarios carrier aggregation clearly outperforms load balancing handovers and HSPA+LTE aggregation simply brings the same benefits to inter-RAT domain. In uplink the aggregation (both in intra- and inter-RAT domain) is however less appealing due to UL coverage and UE power consumption limitations. In addition to the data rate gains, HSPA+LTE aggregation allows more relaxed re-farming strategies for HSPA spectrum; HSPA+LTE aggregation capable UEs can enjoy improved data rates by utilizing efficiently both LTE and HSPA spectrum without reducing the data rates of the HSPA UEs.
Figure 12: Average downlink data rate before and after refarming of one HSPA carrier (assuming low-to-medium system loading, 10MHz LTE and 2x5MHz HSPA before refarming)
Figure 12 illustrates downlink data rates in a single UE scenario, where before re-farming both HSPA and LTE have 10MHz bandwidth. By re-farming of one HSPA carrier the data rates of LTE UEs can be boosted by ~50% but that happens at the cost of ~50% lower HSPA data rates. With HSPA+LTE aggregation it is possible to postpone the re-farming until HSPA penetration is very low, but at the same time provide almost 100% higher data rate for the LTE UEs with HSPA+LTE carrier aggregation capability.
Figure 13: Potential split/merger points of HSPA + LTE aggregation shown on top of current network architecture
In the next sections each of these potential split/merge points are analyzed in more detail. It is worth to notice that no changes to the existing system architecture or protocols regarding the operation (HSPA only or LTE only operation) are envisioned. Thus, the following architecture considerations focus purely on the HSPA+LTE carrier aggregation operation. Table 6 summarizes some aspects of the different architecture options.
Table 6: A high level summary of the different architecture approachses for HSPA+LTE aggregation
RAN aspects
Minimal impact on RAN
UE aspects
Simultaneous UL on both RATs required Impact on battery
Other aspects
Challenges in optimizing usage of resources
HSPA RNC
No/minimal changes at NodeB Changes at RNC and eNB to support new interface between eNB and RNC
RNC based scheduling slower than base station based one Cannot benefit from the faster setups over LTE All traffic go through 3G CN
HSPA NodeB
Most changes limited to base station No impact on core network No or limited impact on higher layers Changes to radio interface needed to transport signaling and setup of radio interfaces Most changes limited to base station No impact on core network No or limited impact on higher layers Changes to radio interface needed to transport of signaling and setup of radio interfaces
UL range reduction if simultaneous HSPA/LTE UL For HSPA UL only case, HARQ timing may be an issue
Very good performance due to fast scheduling and shallow reordering Fast load balancing Cannot benefit from the faster setups over LTE All traffic go through 3G CN
LTE eNB
Very good performance due to fast scheduling and shallow reordering Fast load balancing
Having the split/merger at service level or CN level has the following advantage: It can be introduced with a minimum impact on the RAN. In principle, no changes are required to the user plane processing; however to make it feasible in practice (i.e. to mitigate the disadvantages listed below), some changes in the UE will be required, and the network side of the RRC layers need to be aware of the dual-radio operation.
Both alternatives would however require pure dual-radio with simultaneous dual-transmission which leads (at least) to the following further requirements and drawbacks: o o o o o Double security, mobility context and CN protocol layers UE total maximum TX power management and handling SAR requirements (e.g., UE TX power reduced by 3 dB in both systems) Significant impact to UE battery life due to needing to operate multiple power amplifiers simultaneously UE RF implementation issues such as inter-modulation, or interference to own receiver, due to two simultaneous transmissions Challenging to optimise usage of HSPA or LTE resources, leading to that the full capacity gain cannot be achieved.
In addition it appears that the RAN control plane processing could not stay agnostic to dual-radio user plane due to tight interworking of the HSPA and LTE. At least some level of coordination of the two RRC protocol layers would be inevitable due to access and mobility management.
The disadvantages are: The RNC scheduler will lead to suboptimum performance Deep reordering required in the UE, due to potentially very different delays in the two RATs Simultaneous uplink transmission on LTE and HSPA may be required.
Introducing the data split/merger at NodeB level (Figure 17) would limit most of the changes to base station level which could be seen more desirable especially in the context of multi-radio base stations.
While an aggregation in the RNC has the same main disadvantages as the CN/service layer solution, an aggregation in the node B has the possibility to avoid several of the drawbacks. Providing that the node B eNode B interface has low enough latency, the following advantages may be achievable: Very good performance, due to fast scheduling and shallow reordering Possibility to use either HSPA uplink or LTE uplink or both for fast feedback No impact on core network nodes, and very limited if any impact to the existing RNC functionalities No or very limited impact to higher layers
Both HSPA and LTE uplink could be used simultaneously for fast feedback. This approach has the advantage of requiring little or no layer 1 change. The drawback is that UL range is reduced due to simultaneous transmission on both links. A single uplink may be desirable to avoid UL range reduction, though the approach has the following disadvantages: Standardization changes required to lower layers, in particular for the uplink layer 1 control signalling. This is true irrespective of which RAT is used for uplink transmission. LTE data would have to be routed via RNC which would violate the flat architecture design principle of LTE and lead to increased RNC load (the LTE UEs could be however still served using the existing LTE architecture without routing their traffic via RNC).
Here, the most appealing alternative would be to reuse the PDCP, RLC and MAC-d protocols from HSPA. With this approach, the impact is limited to MAC in LTE. As the control plane and the data routing to higher layers in this architecture would be managed by UTRAN, the HSPA uplink might be seen as most natural choice in case that single UL is desired. In this case, also the LTE feedback (HARQ acknowledgements and CQI) would have to be transmitted via HSPA uplink which can be challenging due to the shorter (1ms) TTI of LTE. This approach is depicted in Figure 18. Alternatively the uplink control and data could be mapped on the LTE. This approach might be more attractive from the control signalling perspective, but would require additional user plane modifications.
Figure 18: Single uplink with data split/merger at NodeB. Most of the protocols are used as is: the main impact is seen in LTE MAC
Since in HSPA the PDCP and RLC layers are located in RNC, the data split should take place at/below LTE RLC layer. Using the LTE MAC could potentially lead to a more optimized performance and flexibility, but it would require rather dramatic modifications in the LTE MAC implementation, particularly in the UE, as major parts of the MAC-ehs would have to be ported to the LTE MAC to support HSDPA L1. If the data split is however performed in the RLC-MAC interface both the LTE and HSPA MAC (and L1) can be kept intact (if so desired) making this the most appealing alternative from the implementation complexity point of view. As a consequence of the proposed architecture choices described above, the RLC, PDCP, and RRC protocol layers of HSPA side would not be used for HSPA+LTE aggregation, instead only the LTE RLC, PDCP, and RRC would be utilized. Similarly, as S1 interface is terminated in LTE eNodeB, the GPRS packet core protocols are not utilized, but core network functions are provided by EPC. As mentioned in Section 4.2, both HSPA and LTE uplink can be used simultaneously for fast feedback to avoid any layer 1 change, but this approach comes with a drawback of UL range reduction. Alternatively, if a single UL is used to maintain the UL range, the HSPA feedback (CQI, HARQ status) would have to be delivered via LTE UL, as illustrated in Figure 20. The tight delay budget for delivering such feedback 28 4G Americas HSPA+LTE Carrier Aggregation June 2012
implies that NodeB and eNodeB should either be co-located or integrated into one multi-radio BTS. This is however not that strict requirement as co-location is in any case desirable to maximize the overlapping of the cell coverage areas, as well as to minimize the site costs. This approach is also well aligned with the LTE-Advanced carrier aggregation frame work where single UL can provide feedback for multiple DL carriers.
As a summary, the solution with the split in the eNode B shares many of benefits of having the split in the node B: o o o No impacts at core network or services level, and only minor impacts on RNC Data split at BTS level enables fast load balancing Single uplink via LTE UL possible thus maximizing the uplink range
In addition, using the eNode B as the aggregation point also means that o o The LTE data flow would not have to travel via RNC Allows to utilize the existing LTE CA framework
Quite naturally, the solution also has these disadvantages: Standardization changes required to lower layers, in particular for the uplink layer 1 control signalling, if a single uplink is used for feedback of both HSPA and LTE. Since the RNC maintains the overall responsibility for the HSPA resources, the eNodeB cannot control all the resources in the node B, leading to somewhat degraded performance gains.
5. PRACTICAL IMPLEMENTATION ASPECTS OF HSPA+LTE AGGREGATION 5.1 BASE STATION RADIO IMPLEMENTATION ASPECTS
Figure 21 shows a very simplified block-diagram of a base station capable of transmitting on two frequency bands with two transmit antennas common to the bands. The same transmit chain can in principle be able to transmit either LTE, HSPA, or even both LTE and HSPA carriers simultaneously on separate carrier frequencies within the bandwidth of the transmitter, e.g. a 10 MHz transmitter could support one 10 MHz LTE carrier or two adjacent 5 MHz HSPA carriers, and a 40 MHz transmitter could support two adjacent 20 MHz carriers or 8 adjacent 5 MHz HSPA carriers, or even one 20 MHz LTE carrier and next to it 4 adjacent 5 MHz HSPA carriers.
PA
Filter
Filter
Upconversion
Filter
DAC
Filter Band A
Filter
Filter
DAC
Filter
Filter
Upconversion
Filter
DAC
PA
Filter Band B
Filter
Filter
DAC
As a concrete example, we can consider a co-sited HSPA and LTE deployment with one or several HSPA carriers on PCS band and one LTE carrier on AWS band. Extending such deployment to support HSPA+LTE aggregation would be directly able to utilize the RF hardware already in place. This can be generalized to say that an existing co-sited deployment of HSPA and LTE RATs can be extended to support HSPA+LTE aggregation without any new requirements to the already deployed RF hardware. Note that new baseband functionality needs to be introduced. The architectures with data split point in the base station (HSPA Node B or LTE eNode B) would be easiest to implement with a multi-standard radio base station, where both RATs are served by the same physical entity. If the uplink is only limited to one or the other RAT, then a fast feedback loop would be required from one RAT to the other to get the uplink channel state information and HARQ ACK/NACK feedback across to the other RAT. These architectures assume either a high-speed, low latency interface between two base stations, or more advantageously one multi-standard radio base station.
PA
PA
Amp
Amp
Filter
Filter
Downconversion
ADC
Amp
Amp
Band A
Filter
Filter
Downconversion
ADC
Filter Band B
Filter
ADC
A device supporting intra-band carrier aggregation could use a single wide band receiver rather than two narrow band ones, e.g. a Dual Cell HSDPA UE can be expected to have one 10 MHz receiver rather than two 5 MHz receivers. If the aggregated carriers can be non-adjacent, then there may be a need to go to the architecture similar to that used in inter-band carrier aggregation.
UE receiver BB processing
Filter
Filter
ADC
Amp Amp
Amp Amp
31
Figure 23 shows a very simplified block-diagram of a UE capable of transmitting on two frequency bands, but not simultaneously. Such dual-band (or multi-band) transmitter architectures are expected to be more commonplace than those capable of using multiple transmitter chains simultaneously, e.g. a 10 MHz transmitter could be used for aggregating two adjacent 5 MHz HSUPA carriers for Dual Cell HSUPA configuration, or be able to transmit on one 10 MHz LTE carrier. It however is to be noted that this architecture is not readily able to lend itself to transmitting on multiple frequency bands a full dualtransmit chain architecture would be required for that.
Filter Band B
Filter
upconversion
Figure 24 shows a very simplified block-diagram of a UE capable of transmitting on two frequency bands simultaneously, i.e. capable of uplink carrier aggregation on two bands. Such dual-band (or multi-band) transmitter able to lend itself to transmitting on multiple frequency bands and thus would be able to support also HSPA+LTE aggregation in the uplink.
Band A
PA
upconversion
Filter Band B
Filter
upconversion
Filter
DAC
BB processing
Filter
Filter
Filter
DAC
PA
BB processing
PA
PA
6. CONCLUSION
HSPA+ and LTE are the overwhelming mobile broadband technologies of choice for operators throughout the world. The evolution of both HSPA+ and LTE standards has introduced aggregation of carriers for higher data rates, better load balancing and increased spectrum utilization, and since the dawn of LTE, the standard support for radio level interworking for HSPA and LTE radios has been included. A natural continuation of such development is to tighten the interworking even further and introduce similar aggregation of carriers between the two radio access technologies. The same gain mechanisms that have been seen beneficial for Multicarrier HSDPA as well as LTE Carrier Aggregation can be benefited from by aggregating HSPA with LTE. At low or medium load, HSPA+LTE aggregation is able to take advantage of the unused resources leading to significant data rate increases both at the cell edge and the cell center for the carrier aggregation capable devices. In addition, the carrier aggregation enables fast (millisecond level) load balancing across the carriers thus improving the data rates of all users. A number of possible network architectures can be foreseen for HSPA+LTE aggregation, and are briefly touched upon in this white paper. Most promising architecture options are seen with co-located multiradio base stations with the base station (NodeB + eNodeB) acting as the data aggregation point, and simultaneously maintaining the existing network architecture for the devices connecting to the network with one radio system at a time only. This architecture can utilize some of the already deployed RF hardware in the base station, whereas new baseband functionality managing the data flow will be needed. On the device side receiver radio architectures capable of multiband carrier aggregation should be suitable also for aggregated HSPA+LTE. While Dual-Cell HSDPA is already in commercial operation, and higher levels of HSPA carrier aggregation as well as LTE carrier aggregation are part of 3GPP specifications existing today, HSPA+LTE aggregation is currently not standardized. Although conceptually straightforward and building on already standardized concepts, HSPA+LTE aggregation is a major feature, with a standardization effort comparable to that of LTE carrier aggregation.
ABBREVIATIONS
3GPP 4C-HSDPA 8C-HSDPA A/N ACK ADC ASN.1 AWS BB BTS CA CC CDMA CIF CN CQI DAC DB DC-HSDPA DC-HSUPA DCH DL E-DCH E-UTRAN EPC FACH GPRS Gbps GSM HS-SCCH HARQ HSDPA HSPA HSUPA ID ITU L1 LTE LTE-A MAC MAC-d MAC-hs MAC-ehs Mbps MHz MIMO 3rd Generation Partnership Project 4-Carrier HSDPA 8-Carrier HSDPA ACK/NACK Acknowledgement Analog to Digital Conversion Abstract Syntax Notation One Advanced Wireless Spectrum Base Band Base Transceiver Station Carrier Aggregation Component Carrier Code Division Multiple Access Carrier Indicator Field Core Network Channel Quality Indication Digital to Analog Conversion Dual Band Dual Cell HSDPA Dual Cell HSUPA Dedicated Channel Downlink Enhanced DCH Evolved UTRAN Evolved Packet Core Forward Access Cannel General Packet Radio Service Gigabits per second Global System for Mobile Communications High Speed Shared Control Channel Hybrid Automatic Repeat reQuest High Speed Downlink Packet Access High Speed Packet Access High Speed Uplink Packet Access Identity International Telecommunication Union Layer one Long Term Evolution LTE Advanced Medium Access Control MAC dedicated MAC high speed MAC enhanced high speed Megabits per second Megahertz Multiple Input Multiple Output
NACK O&M OEM PA PCC PCell PCH PCS PDCCH PDSCH PHY PDCP QAM RAT Rel RF RLC RNC ROHC RRC SCC SCell SIB SMR SON TTI TX UE UL UMTS URA US UTRAN WCDMA WRC
Negative ACK Operation and Maintenance Original Equipment Manufacturer Power Amplifier Primary Component Carrier Primary Serving Cell Paging Channel Personal Communications Service Physical Downlink Control Channel Physical Downlink Shared Channel Physical [layer] Packet Data Convergence Protocol Quadrature Amplitude Modulation Radio Access Technology Release Radio Frequency Radio Link Control Radio Network Controller Robust Header Compression Radio Resource Control Secondary Component Carrier Secondary Serving Cell System Information Block Specialized Mobile Radio Self Optimizing Network Transmission Time Interval Transmit User Equipment Uplink Universal Mobile Telecommunications System UMTS Routing Area United States UMTS Terrestrial Radio Access Network Wideband CDMA World Radio Congress
ACKNOWLEDGEMENTS
The mission of 4G Americas is to promote, facilitate and advocate for the deployment and adoption of the 3GPP family of technologies throughout the Americas. 4G Americas' Board of Governor members include Alcatel-Lucent, Amrica Mvil, AT&T, Cable & Wireless, CommScope, Entel, Ericsson, Gemalto, HP, Huawei, Nokia Siemens Networks, Openwave, Powerwave, Qualcomm, Research In Motion (RIM), Rogers, T-Mobile USA and Telefnica. 4G Americas would like to recognize the project leadership and important contributions of Karri Rantaaho of Nokia Siemens Networks (NSN), as well as representatives from the other member companies on 4G Americas Board of Governors who participated in the development of this white paper.