OFDMA vs SC-FDMA

The multiple access of the radio channel in LTE is based on OFDMA (Orthogonal Frequency Division Multiple Access) in the DL and SC-FDMA (Single-Carrier Frequency Division Multiple Access) in the UL. These two techniques have a number of similarities but also some major differences as the UL and DL have different issues to cope with.

Both multiple access techniques aim at spreading the narrowband data signal over a higher bandwidth to provide high data rates and cope with the frequency selective radio channels. In order to achieve that the wideband channel is split into a number of narrowband subcarriers and each user is allocated subcarriers which are not fading for the user’s particular radio channel. The data for this user is then transmitted using these subcarriers only. The main advantage of OFDMA and SC-FDMA over other multicarrier techniques is that the subcarriers are orthogonal to each other allowing higher spectrum efficiency.

Let's first analyse the OFDMA technique which is a bit more standard than the SC-FDMA. The signal processing chain is shown in the next figure:




Each data symbol (QAM symbol) modulates a single narrowband subcarrier (15kHz). All these subcarriers are orthogonal to each other in order to avoid interference. All the modulated symbols are transmitted simultaneously over the air.



Now the superposition of all the modulated subcarriers is a noise-like signal according to the central limit theorem. Thus the variation of the signal amplitude is quite high leading to a high Peak-to-Average-Ratio (PAPR). This requires expensive power amplifiers and high power transmission which is acceptable for the eNB but not desirable for the UE.

Solution to this problem is the introduction of SC-FDMA which is a similar technique offering the same advantages as OFDMA without the high PAPR issue. In SC-FDMA each data symbol modulates the whole used wideband carrier instead of a narrowband one and the modulated symbols are transmitted sequentially over air. Thus the final transmitted signal is a single carrier one unlike OFDMA where the final signal is the superposition of a great amount of carriers.





It is obvious from the figures that the two techniques transmit the same amount of data symbols in the same time period and using the same bandwidth. However in SC-FDMA the UE needs to transmit only one carrier (wideband) at a time containing the information of one data symbol, while in OFDMA a number of subcarriers (narrowband) need to be transmitted at each time period. Thus the SC-FDMA technique is more power efficient.

Email the author

Mobility Management in LTE/EPS

Mobility Management is the function of controlling the mobility (and security) of the UE when using the RAN while the Session Management is the function of providing IP connectivity (and resources) to the UE. Connection Management is a subfunction of Mobility Management controlling the connection of the UE to the network. These functions are implemented by protocols belonging to the NAS (Non-Access Stratum).



Mobility Management in LTE/EPS known as EPS Mobility Management (EMM) is similar to the UMTS one with some changes stemming mainly from the LTE requirement for “Always On” IP connectivity. This requirement is achieved by providing an EPS Bearer (Default EPS Bearer) to the UE once it attaches to the network and by keeping it until it switches off. Thus the Mobility and Session Management mechanisms in LTE are interconnected. For more info on the EPS Bearer look here.

Mobility Management (EMM)

The main EMM procedures are as follows:

• Attach
  
• Detach
  
• Tracking Area update
  
• Paging
  
• Identification
  
• Security Mode Control

The EMM defines two main states for the UE, while other intermediate states are also defined:

EMM-Deregistered
When the UE is not attached to the network and thus the network doesn’t know its location and cannot reach it.

EMM-Registered
When the UE is attached to the network and an EPS Bearer is established for it.

Independently of the EMM states there are two EPS Connection Management (ECM) states the UE can be in:

ECM-Idle
When there is no NAS connection established for the UE and no physical resources are allocated to it (i.e. Radio and S1 Bearers). It can still exchange data as its EPS Bearer still exists but a NAS connection needs to be established first in order to get physical resources allocated.

ECM-Connected
When there is a NAS connection established for the UE and physical resources are allocated to it. In order to get to the EMM-Registered state the UE needs to get to the ECM-Connected mode first.

The NAS connection which changes the ECM-Idle to the ECM-Connected mode is established when the UE needs to send an initial NAS message such as Attach/Detach Request, Tracking Area Update, etc. Then an RRC connection needs to be established between the UE and the eNodeB and in sequence an S1_MME connection. After that the NAS connection between the UE and the MME is established. The initial NAS message can then be sent by the UE. In the same manner the NAS connection release which brings the UE back to ECM-Idle mode takes place after the RRC connection and the S1_MME connections are released.

Session Management (ESM)

The main EPS Session Management (ESM) procedures which realize the bearer handling and IP connectivity are as follows:

• Default EPS Bearer Context Activation
  
• EPS Bearer Context Deactivation
  
• Dedicated EPS Bearer Context Activation
  
• EPS Bearer Context Modification
  
• UE Requested PDN Connectivity
  
• UE Requested Disconnect
  
• UE Requested Bearer Resource Allocation
  
• UE Requested Bearer Resource Modification

All these procedures can be performed only after a NAS connection is established (ECM-Connected). The Default EPS Bearer setup is performed during the EMM Attach procedure.

ESM defines two main states for the UE:

Bearer Context Inactive
When there is no EPS Bearer for the UE.

Bearer Context Active
When there is at least EPS Bearer for the UE. An EPS Bearer Context can exist even if no physical resources are allocated to the UE (i.e. Radio and S1 Bearers).

Email the author

Detailed Description of the QoS Mechanism in EPS

A Packet Filter has to be set up at the PDN GW (and signaled to the UE) for each SDF or SDF Aggregate in order to allow the correct mapping of data to the EPS Bearer and effectively the correct routing. The EPS Bearer is associated with one TFT (one in UL and one in DL) and thus an EPS Bearer can carry only one SDF or only one SDF Aggregate while all data of the same EPS Bearer will experience the same QoS treatment meaning. The SDF can be mapped to an EPS Bearer only if they have the same QCI and ARP.

The Packet Filters are sequentially applied to the incoming data (on the UE in UL and PDN GW in DL) according to the Evaluation-Precedence-Index values of the Packet Filters and eventually the SDF (if one Packet Filter exists for the TFT) or the SDF Aggregate (If more Packet Filters exist for the TFT) is mapped to a TFT and thus to an EPS Bearer.

In UL the UE creates a binding between the SDF or SDF Aggregate and the Radio Bearer (RB) carrying it, the eNB creates a binding between the RB and S1 Bearer, the SGW creates a binding between the S1 Bearer and the S5/S8 Bearer. In the DL the PDN GW creates a binding between the SDF or SDF Aggregate and the S5/S8 Bearer carrying it, while the rest mappings are as in UL.

Any non-matching data should be sent to the Bearer which has no Packet Filters associated. If no such Bearer exists, data shall be discarded.

Traffic Flow Templates
A TFT (Traffic Flow Template) consists of one or multiple Packet Filters (1-8) and is used to discriminate data packets from different applications, with different QoS requirements, etc. in order to route them appropriately. Each Packet Filter is aimed at isolating the data from one application protocol (e.g. FTP), so effectively a TFT can carry data from more than one application protocol, which have to share the same QoS characteristics though. The TFT’s are pre-configured at the PDN GW.

Packet Filters
The Packet Filter has a unique packet identifier (1-8) within the TFT and consists of one or more of the following attributes depending on its configuration with regards to the application to be carried:

• Source/Destination address with subnet mask
  IP address with subnet mask; source is valid on DL and destination on UL
  
• Protocol number
  Number of higher protocol (e.g. TCP/UDP)
  
• Destination port range
  Port range of the application (e.g. HTTP)
  
• Source port range
  Port range of the application (e.g. HTTP)
  
• IPsec security parameter index
  Arbitrary number between 256 -16639 to identify the secure connection between two entities
  
• Type of service
  Identifies the QoS
  
• Flow level
  IPv6 flow level. Not used in IPv4.

Each Packet Filter has an Evaluation-Precedence-Index which is unique across all TFT’s of the same APN which indicates the priority in which the Packet Filters will be applied to the packets. Highest priority is 0 and lowest is 255. Packet filters are signaled to the UE during NAS procedures.

TFT Example

A PDN GW has 2 different TFT’s stored as follows. The first one is aimed at transferring VoIP, with two different options (over TCP or UDP) and the second one is aimed for FTP services. The PDN GW checks first the two packet filters of the VoIP TFT and then the FTP one.

When data from an FTP service reaches the PDN GW, the PDN GW applies Packet Filter 1 first, then Packet Filter 2 and finally Packet Filter 3 as per their Evaluation-Precedence-Index values. Packet Filter 3 matches its attributes (protocol number and ports) and the PDN GW creates a binding between the FTP SDF and the TFT of Packet Filter 3 (i.e. FTP TFT).Then the PDN GW forwards the data to the EPS Bearer which is associated with that TFT. The data will then follow the appropriate S5/S8, S1 and finally Radio Bearer which constitute that EPS Bearer.

For a description of the EPS Bearer and the QoS Model in EPS click here.

Email the author

QoS in LTE/EPS

EPS Bearer Architecture

EPS introduces an E2E QoS model based on the EPS Bearer, which replaces the PDP Context of GPRS. The EPS Bearer is a logical link between the UE and the PDN GW providing a specific QoS along the whole path. The EPS Bearer Architecture consists of multiple layers as in UMTS and the service of each layer is provided through the service of the layer immediately below it. Thus the required QoS to the end user is provided by each Bearer across all layers. The E2E Service is completed by an external Bearer from the edge of PLMN to the final external destination and not controlled by the PLMN.

Bearer Level QoS Paremeters

The QoS of an EPS Bearer is defined by the following QoS parameters:

• QCI (Quality Class Identifier)
  QCI is an index referring to a number of different sets of minimum QoS characteristics, such as priority, delay, etc. required by a service. The different QCI’s can be achieved by different packet forwarding treatment (e.g. scheduling, queue management, RLC config, etc.) at the network nodes. The network has to be pre-configured to provide the supported QCI’s. There are 9 standardised QCI’s and associated with example services. The characteristics of the QCI are:

ARP (Allocation and Retention Priority)
It is a priority indicator in order to allow the network to reject the establishment or modification of new Bearers or discard existing ones in cases of limited resources. After the Bearer establishment, it does not affect routing.

GBR (Guaranteed Bit Rate)
Applies to Bearers with Resource Type = GBR and indicates the minimum bitrate to be provided for this service.

MBR (Max Bit Rate)
Applies to Bearers with Resource Type = GBR and indicates the maximum bitrate to be provided for this service. GBR=MBR for now.

APN - AMBR (APN – Aggregated MBR)
Applies to Bearers with Resource Type = non-GBR. It is the maximum bitrate allowed across all Bearers of the same UE for each UE-APN connection and it is stored in HSS. One non-GBR Bearer can have the whole capacity if all other non-GBR are zero.

UE - AMBR (UE – Aggregated MBR)
Applies to Bearers with Resource Type = non-GBR. It is the maximum bitrate allowed of the same UE and it is stored in HSS. One non-GBR Bearer can have the whole capacity if all other non-GBR are zero. UE-AMBR = sum(APN-AMBR) for all APN connections of the UE.

QoS Mechanism
When an UE is attaching to the network, an EPS Bearer is always set up and lasts until the UE detaches in order to provide an ‘Always On’ IP connectivity, reducing set up latency and excessive signaling. This Bearer is the Default Bearer and its QoS parameters are set by the network based on the UE subscription profile stored in the HSS. The Default Bearer is always non-GBR. More EPS Bearers can be established for one UE, known as Dedicated Bearers. The Dedicated Bearers are modified or created always after network trigger. The GBR Bearers have fixed allocated dedicated network resources related to the GBR, while non-GBR not.

In order to map the incoming data packets to the correct EPS Bearer, the packets need to have a QoS requirement which will then be satisfied by the Bearer. All the data packets from one application protocol (e.g. FTP) within the same IP-CAN session constitute an SDF (Service Data Flow) and the SDF has a QoS requirement as defined by its QoS parameters, described before: QCI, ARP, MBR, and GBR. A set of SDF’s within the same IP-CAN session with the same QCI and ARP constitutes an SDF Aggregate which can be treated in the same way as a single SDF. One EPS Bearer can then carry all the SDF’s which have exactly the same QoS parameters as the Bearer. Thus an EPS Bearer carries the data from one application or from multiple applications which share exactly the same QoS requirements.

The exact mechanism for the mapping of SDF’s to EPS Bearers is using the TFT concept and is described in detail here.

Email the author

Do We Actually Need LTE??

Everyone involved somehow in the mobile technologies and market knows very well that the hot topic of the industry is LTE (Long Term Evolution of UMTS) or EPS (Evolved Packet System) as is the correct term as per 3GPP standards.

From a technology point of view it is easy to understand that EPS is by far better than 2G and 3G in order to support the current and future packet-centric services. However a lot of the industry thought leaders remain sceptical towards EPS as they claim that the return on investment of such a dramatic change is low for the operators. The cost of deploying EPS is quite high as new spectrum licence has to be acquired, new equipment in the RAN and Core need to be purchased and new IP-based transmission network needs to be deployed. While on the other hand the data services haven't really taken off yet in most of the countries and the voice and SMS services remain the main source of income for most mobile operators.

I think the answer to this question came to me after I got my iPhone. The use of data services really EXPLODES once someone gets one of the new iPhone-like smartphone consuming on average 30 times more bandwidth than a voice user! Even people who never used to do packet services before, become regular data users with the smartphones. And the experience from UK and US has shown that once these devices become available people start getting them in bulk exploding the bandwidth usage from one day to another.

One can now claim that HSPA is fast enough to provide the data rates required by most of the current applications which are not that demanding (email, browsing, IM, facebook, etc.). However the main killer is not the air interface, but the backhaul (see also The Future of Mobile Backhaul for details). Operators need to increase the backhaul bandwidth significantly and the cost for that is not trivial as they need to purchase a huge amount of leased lines in the current TDM/ATM based deployments. Moving to IP is the way forward but this what EPS is all about! Another killer is signaling. Smartphones generate so much signaling in the current voice-centric networks that make it really inefficient for the capacity but also for the mobile's battery life.

Moving a bit away from the immediate problems that operators are facing now and thinking longer term, we will see that EPS will allow the support of completely new services such as online gaming, mobile TV, etc. which are now very difficult to get with current networks. These services may seem very distant for most of the people, but we shouldn't underestimate the fact that young people and kids or the "internet generation" as we call them adapt very fast to the new fancy services and change their consuming behaviours very quickly. Facebook is a good example, which has almost become the killer application that 3G operators were looking for a long time.

Finally if EPS can provide the data rates that is promising, then mobile operators can drastically enter the broadband market and compete with the fixed line operators making significant amount of business in this field before WiMAX operators do so. Why should I have a fixed broadband service if I can get all my telecoms services with one contract? Furthermore they could also provide mobile broadband to countries with poor infrastracture and become the main broadband providers as their implementation costs will be lower than the fixed ones.

Email the author