3G封包存取的有效频谱技术 (英)

Spectral Efficient Technologies in 3G for Packet AccessBhanu Srinivas Valluri

The increasing availability of a broad range of new high-speed data services is fuelling demand for more bandwidth in order to improve the user experience, especially over mobile networks. This article explains the available and upcoming spectral efficient technologies in 3G for packet data, which includes CDMA 2000 based 1xEVDO and WCDMA based HSDPA.

1xEV-DO (Evolution-Data Optimized) is the commercially available 3G technology that delivers data rates exceeding 2 Mbps in 1.25 MHz of spectrum, offering bandwidth efficiency for data traffic that is three to four times greater than other voice-centric standards such as CDMA2000 1xRTT and WCDMA.

HSDPA is a UMTS packet air interface (add-on solution on top of 3GPP R99/R4 architecture) that allows up to 3.6 Mbps peak data rate for a category 6 mobile per user with a classical rake receiver and up to 14.4 Mbps peak data rate for a category 10 mobile per user with advanced receiver solutions.

These new technologies offer mobile operators significantly improved data speeds, increased capacity and a very real competitive edge against wired/wireless broadband providers.

Both HSDPA and 1xEV-DO, enhance downlink packet data performances. HSDPA and 1xEV-DO are based on the same set of technologies to improve spectral efficiency for data services—such as shared downlink packet data channel and high peak data rates—using high-order modulation and adaptive modulation and coding, HARQ retransmission schemes, fast scheduling and shorter frame sizes.

Figure 1 depicts the migration of HSDPA and CDMA 1xEV-DO standards, and Figure 2 displays the performance evaluation and data rates of these cellular technologies.

Figure 1:  HSDPA & CDMA 1xEV-DO Standards Migration

Figure 2:  Performance evolution of cellular technologies

Architecture: CDMA 1X Ev-Do Overview

CDMA2000 1xEV-DO (Evolution-Data Optimized) is the commercially available 3G technology that delivers data rates exceeding 2 Mbps in 1.25 MHz of spectrum, offering bandwidth efficiency for data traffic that is three to four times greater than other voice-centric standards such as CDMA2000 1xRTT and W-CDMA.

Even though 1xEV-DO is part of the CDMA2000 standard, it does not rely on any element in the CDMA voice network to provide service, mobility or roaming. The operator does not require a Mobile Switching Center (MSC) or network elements such as the home and visitor location registers (HLR/VLR). As a result, 1xEV-DO can be deployed by any voice operator irrespective of the voice technology it is currently using. All that is required for launching 1xEV-DO service is 1.25 MHz of paired spectrum.

Understanding 1xEV-DO

1xEV-DO Network Architecture
A 1xEV-DO network (Figure 3) has three key elements:

Radio Nodes (RNs) Radio Network Controller (RNC) Packet Data Serving Node (PDSN)

Figure 3:  1xEV-DO Network Architecture

Each radio node typically supports three sectors and serves one cell. A dedicated transceiver in each sector terminates the 1xEV-DO airlink between the subscriber modem and the radio node. Higher layers of the 1xEV-DO protocol are processed at the RNC. The RNC also manages handoffs and passes user data between the RNs and the PDSN. The PDSN is a wireless edge router that connects the radio network to the Internet. Unlike some other 3G wireless technologies, this architecture does not depend upon a Mobile Switching Center (MSC).

In addition to the RNC and the PDSN, a 1xEV-DO data center has an aggregation router, an element management system (EMS) and several ISP servers. The aggregation router terminates IP traffic from the RNs and passes it to the RNC. The EMS manages the radio access network. Commonly used ISP servers include, among others, standard IP servers for the Domain Name System (DNS), Dynamic Host Configuration Protocol (DHCP) and Authentication, Authorization, and Accounting (AAA).

1xEV-DO Forward Link (Downlink)
The basic unit of transmission on the 1xEV-DO forward link is the airlink frame. Airlink frames destined for different users in the same sector are time-division multiplexed. The radio nodes transmit each airlink frame at the link rate, which can vary between 38.4 Kbps and 2.4 Mbps. The link rate depends on the signal-to-interference-plus-noise-ratio (SINR) at the subscriber’s location. SINR can vary significantly within a cell (Figure 4). This variation is an inherent characteristic of all wireless systems and occurs primarily because of variations in RF propagation loss, building penetration loss, fading effects, and co-channel interference. As a result, the link rate experienced by a subscriber depends on the position of his or her home (within the cell)—just as it does in DSL.

Figure 4:  Link Rate Distribution in a 1xEV-DO Cell

SINR = S/(N + I), where S is the received signal power, I is the interference received from other sectors and N is the receiver’s thermal noise.

As in cable modems, since the link between the network and the subscribers is shared, depending on system load, the actual data rate experienced by a subscriber can be less than the link rate. Roughly speaking, the number of subscribers who are receiving data at the same time in the same sector determines the system load. For instance, when two subscribers are downloading simultaneously at a fixed link rate of 1.2 Mbps, the actual data rate they experience will be around 600 kbps.

1xEV-DO Reverse Link (Uplink)
1xEV-DO subscribers can transmit on the reverse link at data rates ranging from 9.6 kbps to 153.6 kbps. Unlike the forward link, in which a scheduler time-division multiplexes airlink frames over the channel, the reverse link uses CDMA, which allows multiple users to transmit at the same time. As a result there is no difference between a user’s link rate and data rate on the reverse link.

The reverse link sector throughput represents the total link capacity of the reverse link. Computed as the sum of the reverse link data rates of all simultaneously transmitting subscriber modems, it is about 250 kbps per sector or one-fifth the forward link sector throughput.1xEV-DO radio nodes equitably share this throughput among all active subscribers by using a rate control mechanism.

The below section examines four key questions that determine the economics of the service:

How much spectrum is required? How much area can each cell site cover (coverage)? How many subscribers can each cell site support for a given amount of spectrum (capacity)? What are the backhaul and Internet interconnection requirements?

Designing a 1xEV-DO Network

Spectrum Requirements
Operators require just 1.25 MHz of paired spectrum to start deploying a 1xEV-DO network. This spectrum can be in any frequency band between 450-3500 MHz. Bands currently used for wireless voice systems, such as the PCS (1900 MHz) and Cellular (850 MHz) bands, are especially well suited because they allow an operator to lease or reuse existing CDMA cell sites.

Additional spectrum can be added in 1.25 MHz slices. A 1.25 MHz slice of paired spectrum is often referred to as a carrier or a CDMA channel. The capacity of the network increases linearly with the number of carriers deployed.

Determining Cell Size
The radius of a cell is the maximum distance from the radio node at which the radio node can provide acceptable forward and reverse link service. The cell radius determines the number of radio nodes required to cover an area, which has a direct impact on capital expenditure.

Cell radius can be estimated using link budget analysis. This analysis estimates the maximum RF propagation loss or path loss that is allowable between the subscriber modem and the radio node for acceptable service. Path loss can be translated into cell radius using well-established RF propagation models.

Estimating Capacity
Cell capacity is the maximum number of subscribers that can be served by a radio node during busy hour using one carrier. Each carrier represents 1.25 MHz of paired spectrum. Once a cell is being utilized at its capacity, an operator must either deploy more radio nodes or add more carriers. Statistics indicate that residential Internet users download five times more data than they upload. As discussed in "1xEV-DO Reverse Link", the average forward link sector throughput is higher than the reverse link throughput by the same ratio. Therefore, cell capacity can be estimated by just analyzing forward link capacity.

Forward link capacity depends upon:

Link rate distribution Activity factor: Percentage of subscribers that are active during busy hour Average usage: Average amount of data downloaded by an active subscriber during busy hour Target minimum forward data rate Forward link outage probability: percentage of subscribers who may not receive the target minimum data rate during busy hour.

Backhaul and Internet Connection
IP gives operators the flexibility to choose between several available backhaul options. These include dedicated T1/E1 leased lines, a routed network, wireless backhaul, Metro Ethernet, and Frame Relay. In the scenario where T1/E1 leased lines are used, each cell site connects to the nearest central office over a T1/E1 local access line. To reduce local access line charges, multiple incoming T1/E1 circuits may be combined at a (possibly different) central office and delivered to the data center where the RNC is located.

Overall, 1xEV technology (also known as, High Data Rate or HDR) is a high performance and cost effective Internet solution for consumers and business professionals. It offers high speed, high capacity wireless Internet technology, which is compatible with CDMA networks and optimized for packet data services. 1xEV offers a combination of high performance and economic benefits which is unprecedented in systems capable of portable, mobile, and fixed services. 1xEV achieves this performance with minimal network and spectrum resources, providing a highly spectrally efficient technology.

Architecture: HSDPA Overview

Following in the footsteps of GSM, GPRS, and UMTS, now HSDPA (High Speed Downlink Packet Access) has stepped forward as the latest development in mobile radio technology. The objective of HSDPA is optimization of the UMTS system with respect to data services support. UMTS already offers fast data services, such as high-quality video transmission at 384 kbit/s. Building on that, HSDPA uses new technologies to enable data rates of theoretically up to 14 Mbit/s, as well as to increase the capacity of the mobile radio network. As a result, mobile radio operators can offer their customers even more sophisticated multimedia services. HSDPA is particularly suited to extremely asymmetrical data services, which require significantly higher data rates for the transmission from the network to the UE (downlink) than they do for the transmission from the UE to the network (uplink). In addition to providing the desired data rates and maximizing data throughput, HSDPA is intended to increase the robustness of these data services, which typically exhibit only a low tolerance for errors.

It is a cost-efficient upgrade to current UMTS systems and promises to deliver performance comparable to today’s wireless LAN services, but with the added benefit of mobility and ubiquitous coverage.

Basically, HSDPA is a packet-based data service in W-CDMA downlink with data transmission up to 8-10 Mbps (and 20 Mbps for MIMO systems) over a 5MHz bandwidth in WCDMA downlink (Figure 5). It introduces a new common High Speed Downlink Shared Channel (HS-DSCH) shared by several users. In addition, it introduces enablers for the high-speed transmission at the physical layer like the use of a shorter TTI (2 ms), the use of Adaptive Modulation and Coding, and the use of fast retransmission based on hybrid ARQ (HARQ) techniques. These key mechanisms are located within the UMTS BTS.

Figure 5:  HSDPA Protocol Architecture

The main benefits of HSDPA are:

Improved end-user experience of mobile data services, meaning shorter download times through higher bit rates (14 Mbit/s peak rate) combined with reduced delay and quicker responses when using interactive applications such as mobile office or fast Internet access with support for gaming or audio and video downloads. Enhanced system capacity without additional frequency spectrum, thus significantly lowering the costs for delivering mobile data services.

The technical aspects behind the HSDPA concept include the following:

Shared channel transmission Adaptive Modulation and Coding (AMC) Fast Hybrid Automatic Repeat Request (H-ARQ) Fast cell site selection (FCSS) Short transmission time interval (TTI)

Shared channel transmission
The HSDPA concept introduced few additional physical channels. They are High Speed Physical Downlink Shared Channel (HS-PDSCH) and a dedicated HS-Physical Control Channel (HS-DPCCH).

HS-PDSCH: This channel is both time and code shared between users attached to a Node-B. It is the transport mechanism for additional logical channels; they are HS-Downlink Shared Channel (HS-DSCH) and HS-Shared Control Channel (HS-SCCH).

The HS-DSCH code resources consist of one or more canalization codes with a fixed spreading factor (SF) of 16. At the most 15 such codes can be allocated leaving sufficient room for other required control and data carriers. The available code resources are primarily shared in time domain but it is possible to share the code resources using code multiplexing. When it is both time and code shared, two to four users can share the code resources with the same TTI.

HS-DPCCH: This channel is an uplink channel used to carry the acknowledgement signals to the Node-B for each block. It is also used to indicate the Channel Quality (CQI), which is used for Adaptive Modulation and Coding.

Adaptive Modulation and Coding (AMC)
In present WCDMA networks, fast power control is used for radio link adaptation. This power control is done per slot in WCDMA. Basically link adaptation is required because, in cellular communication systems the SINR of the received signal at the UE (Mobile Equipment is called User Equipment for 3rd generation mobile systems) varies over time by as much as 30-40 dB due to fast fading and geographic location in a particular cell. In order to over come this fading effect and improve the system capacity and peak data rates, the transmitted signal to a particular UE is modified in accordance with the signal variations through a process called link adaptation.

In HSDPA the transmission power is kept constant over the TTI (length of the frame is referred to as Transmit Time Interval) and uses adaptive modulation and coding (AMC) as an alternative method to power control in order to improve the spectral efficiency. HSDPA uses higher order modulation schemes like 16-quadrature amplitude modulation (16QAM) besides QPSK. The modulation to be used is adapted according to the radio channel conditions. QPSK can support 2 bits / symbol where as 16QAM can support 4 bits/ symbol, and hence twice the peak rate capability as compared to QPSK, using the channel bandwidth more efficiently. Different code rates used are 1/4, 1/2, 5/8, 3/4. The Node-B (Base Station) receives the Channel Quality indicator (CQI) report and power measurements on the associated channels. Based on these information it determines the transmission data rate. In HSDPA, users close to the Node-B are generally assigned higher modulation with higher code rates.

Fast Hybrid Automatic Repeat Request (H-ARQ)
The H-ARQ protocol (Figure 6) used for HSDPA is stop and wait (SAW). In SAW the transmitter sends a block of TTI (3 slots) and waits until acknowledge (Figure 7) or negative acknowledge (Figure 8) is received from the UE. In order to utilize the time when it waits for the acknowledgements, N parallel SAW-ARQ processes may be set for a UE, so different processes transmit in separate TTI’s. The value of N is explicitly signaled using 3 bits; hence at the most N can be 8.

Figure 6:  N-channel HARQ

Figure 7:  Packet transmitted and received with acknowledge

Figure 8:  Packet transmitted and received with negative-acknowledge

Figure 9:  Packet re-transmission

The UE requests the retransmission of erroneous data received earlier (Figure 9). Ones the UE receives the 2nd transmission, it combines the information from the original transmission with that of the 2nd transmission before trying to decode the message.

Fast and fair scheduling at Node B
Typically in WCDMA networks the packet scheduling is done at the RNC (radio network connection), but in HSDPA the packet scheduler (medium access layer-hs) is shifted to the Node-B. This makes the packet scheduling decisions almost instantaneous. In addition to this, the TTI length is shortened to 2ms. Hence the scheduling is done very fast as its done every TTI.

A first approach for fair scheduling can be Round-Robin method where every user is served in a sequential manner so all the users get the same average allocation time. However, the requirement of high scheduling rate along with the large AMC availability with the HSDPA concept, where the channel is allocated according to the instantaneous channel conditions. Another popular packet scheduling is proportional fair packet scheduling. Here, the order of service is determined by the highest instantaneous relative channel quality. Since the selection is based on relative conditions, still every user gets approximately the same amount of allocation time depending on its channel condition.

Fast cell site selection (FCSS)
Typically on an average 20-30% of the MS’s are in soft or softer handover condition. Soft handover is a handover between two Node-B’s where as softer hand over is between sectors of a Node-B. So it’s very important to track the active set of Node-B’s connected to a UE for communication. FCSS allows a UE to select the Node-B with the best current transmission characteristics. The advantage of this system is that higher data rates can be achieved at most of the time.

Short transmission time interval (TTI)
The length of the frame is referred to as Transmission Time Interval (TTI). The HS-DSCH, which is added in the HSDPA standard, uses this TTI of 2ms than the Release’99 transport channel TTI. This is done to reduce the round trip time, increases the granularity in the scheduling process and for better tracking of the time varying radio channel. Actually the length of the frame is variable and is selected based on traffic supported and the number of supported users. A typical value is 2ms.

H-ARQ
The AMC uses an appropriate modulation and coding scheme according to the channel conditions. Even after AMC, we may land up with errors in the received packets due to the fact that the channel may vary during the packet is on the fly. An automatic repeat request (ARQ) scheme can be used to recover from these link adaptation errors. When the transmitted packet is received erroneous then the receiver requests the transmitter for the retransmission of that erroneous packet. The basic technique is to use the energy of the previously transmitted signal along with the new retransmitted signal to decode the block. There are two main schemes for H-ARQ, Chase combining and Incremental redundancy.

Chase Combining
It involves the retransmission of the same data packet, which was received with errors. Once the retransmission is received, the receiver combines the soft values of the original signal and the retransmitted signal weighted by the SNR prior to decode the data packet.

Advantages: Each transmission and retransmission can be decoded individually (self-decodable), time diversity gain, may be path diversity gain. Disadvantage: Transmission of the entire packet again which is wastage of bandwidth.

Incremental Redundancy (IR)
Incremental Redundancy is used to get maximum performance out of the available bandwidth. Here the retransmitted block consists of only the correction data to the original data that carries no actual information (Redundancy). The additional redundant information is sent incrementally when the first, second ?n retransmissions are received with errors.

Advantages: Reducing the effective data throughput/ bandwidth of a user and using this for another user. Disadvantage: The systematic bits are only sent in the first transmission and not with the retransmission, which makes the retransmissions non-self, decodable. So, if the first transmission is lost due to large fading effects there is no chance of recovering from this situation.

Partial Incremental Redundancy
The Partial IR is the combination of chase combining and IR. The disadvantage with IR is removed here by adding the systematic bits along with the incremental redundant bits (different puncturing bits) in the retransmissions. This makes both original and retransmitted signals self-decodable.

Thus, all the above-mentioned technologies like HSDPA, 1xEV-DO,W-CDMA are leading to 4G (Figure 10).

Figure 10:  There are many possible technology routes to follow.

Comparison

Technical Features HSDPA 1xEV-DODownlink Frame Size 2 ms TTI 1,25, 2,5, 5, 10 ms Variable Frame SizeChannel Feedback Channel Quality reported at 2 ms rate or 500 Hz C/I feedback at 800 Hz (every 1,25 ms)Data user multiplexing TDM/CDM TDM/CDMAdaptive Modulation and Coding QPSK & 16 QAM Mandatory PSK, 8 PSK and 16 QAMHybrid ARQ Chase or Incremental Redundancy (IR) Asynchroneous Incremental RedundancySpreading Factor SF=16 using UTRA OVSF Channelization Codes Wash Code Length 32Control Channel Approach Dedicated Channel pointing to Shared Channel Common Control Channel
Table 1:  Comparison between 1xEV/DO and HSDPA.

As seen in Table 1 and Figure 11, both technologies have the same spectral efficiency, as they are very similar but HSDPA has higher peak data rates and can fully use the remaining voice bandwidth. In addition, multi-session support is possible with HSDPA, which means the capability to support Voice and Data at the same time.

Figure 11:  Spectrum efficiency between 1xEV/DO and HSDPA

Conclusion

Thus, CDMA2000 1xEV-DO (1x Evolution—Data Optimized) is the first third-generation (3G) mobile air interface standard that offers the data rates required to launch a competitive residential broadband service. With proper network design and planning, operators can use 1xEV-DO to launch a service that has operating profits in the second year of its operation and a payback period of less than five years attractive financial returns by any measure. 1xEV-DO is the 3G air interface that is optimized for packet data, not voice. The 1xEV-DO forward link offers each subscriber the highest supportable data rate at any instant by rapidly adapting modulation and coding schemes to changing radio conditions. Subscribers are efficiently multiplexed over the forward link by a scheduler that determines the sequence in which packets should be transmitted. Connections are set up and torn down rapidly to maximize sector throughput, subscribers an "always-on" experience. 1xEV-DO is a particularly attractive technology for CDMA operators. These operators can deploy 1xEV-DO as a channel card upgrade to their existing CDMA base stations. Further, all 1xEV-DO handsets and modem cards support CDMA 1xRTT and IS-95. As 1xEV-DO subscribers move into 1xRTT coverage, they seamlessly handoff to the lower rate data service. Unlike GSM/W-CDMA, there are no interoperability problems between 1xRTT and 1xEV-DO.

HSDPA is an important ingredient needed to ignite global commerce and to enhance human experience as it will provide a ubiquitous access to Wi-Fi applications without any constraint of hot spots and provide seamless access to every type of broadband service that is already used with ADSL. HSDPA is an extremely cost-effective path to higher data rates and provides more efficient use of valuable spectrum. It enables operators to compete effectively in increasingly converged markets and satisfy the need for enhanced QoS and bandwidth-hungry services in an efficient and cost-effective manner.

Today’s markets are hugely competitive. User expectations are formed by their increasing experience with fixed, cable, LAN and wireless networks. HSDPA makes access to sophisticated mobile IP multimedia services a reality for UMTS carriers.

NOTE: Some of the above data and figures are from Airvananet and some other sources from net.

About the Authors V.Bhanu Srinivas an active member of IEEE has around six years of experience in Telecom switching/Mobile Platforms. His email address is bhanusrinivas@yahoo.com.