1. Overview
  2. Abbreviations
  3. Forward Link
    1. Time Division Multiplexing
    2. Channels
      1. Pilot
      2. Medium Access Control
      3. MAC Index
      4. Traffic
      5. Preamble
      6. Control
      7. Paging
  4. Reverse Link
    1. Overview
    2. Channels
      1. Access
      2. Traffic
      3. Reverse Rate Indicator
      4. Data Rate Control
      5. ACK
  5. Adaptive Modulation
  6. Hybrid ARQ
  7. RL Power_Control & System Stability
  8. FL Scheduling
  9. EVDO Signaling Protocol Stack
  10. Standards and References

Overview

EV-DO is an evolution of the IS-95 CDMA mobile radio communication system and stands for “EVolution Data Optimized”. As is evident from the name, EV-DO was designed for optimizing data traffic rather than the standard speech traffic. Hence, EV-DO offers packetized bearer transport only. Voice traffic can be carried over EV-DO using technologies such as VoIP. Another objective for EV-DO was to minimize the changes required to its precursor, IS-95. Hence for example, EV-DO uses the same 1.25 MHz carriers as IS-95. EV-DO is also designed for asymmetric traffic with the forward link capacity significantly higher than the reverse link capacity.

 

EV-DO is one of the specifications from the CDMA2000 3rd Generation wireless systems project. The 1xEV working group was founded in the 3GPP2 in 2000. The current official name is CDMA2000 HRPD (High Rate Packet Data). EV-DO services include high data rate broadband transmission, streaming media and prioritization, Quality of Service (QoS), multicast services, location based services, etc.

 

Over the years, the EV-DO standard itself has been refined and newer revisions have been published. The first standard is known as EV-DO Rev 0. This was followed by EV-DO Rev A, Rev B and Rev C, work on which is ongoing. This white-paper will focus on Rev 0 only.

 

Abbreviations

ACK         Acknowledgment

AN           Access Node (see also BTS)

ARQ         Automatic Repeat Request

AT            Access Terminal (see also MT)

BER          Bit Error Rate

BTS          Base Transceiver Station

CDMA       Code Division Multiple Access

CSI          Channel State Information

DRC         Data Rate Control

EV-DO      Evolution-Data Optimized

FER          Frame Error Rate

FL            Forward Link

FTC          Forward Traffic Channel

HARQ       Hybrid ARQ

HRPD       High Rate Packet Data

IP             Internet Protocol

MAC         Medium Access Control

MT            Mobile Terminal

LUP          Location Update Protocol

PCT          Power Control Threshold

PSK          Phase Shift Keying

QAM         Quadrature Amplitude Modulation

QoS          Quality of Service

QPSK        Quaternary Phase Shift Keying

RAB           Reverse Activity Bit

RAC           Reverse Activity (sub-)Channel

RL             Reverse Link

RLP           Radio Link Protocol

RPC          Reverse Power Control

RRI           Reverse Rate Indicator

RTC          Reverse Traffic Channel

SINR         Signal To Interference and Noise Ratio

SLP           Signaling Link Protocol

SNP          Signaling Network Protocol

VoIP         Voice Over IP

 

Forward Link

Time Division Multiplexing

In a significant departure from IS-95, on the forward link (FL), EV-DO Rev 0 uses TDMA in tandem with CDMA [Esteves02].

 

The FL is broken up into slots. Each slot is of duration 1-2/3 ms (1.667 ms). A set of 16 slots make up a group called a “frame” (duration: 20-2/3 ms). FL user traffic is rapidly multiplexed amongst the slots (we will come back to the scheduler criteria later in this document). This implies that, EV-DO does not require FL power control and each slot is transmitted at maximum power. This also implies that, since each slot is assigned to a specific Access Terminal (AT) or Mobile Terminal (MT) at run time, each slot needs to be “addressed” to an AT in some way. Note that, the duration of a “frame” is the same as the period of the short PN sequence in IS-95 which is the sector identifier (the same short PN sequence is used for the same purpose in EV-DO).

 

The slot format is as follows:

 

 

From the diagram we see that a slot comprises of 2048 chips. The pilot channel is transmitted twice per slot, (no continuously transmitted pilot as in IS-95) 96 chips each time. In case there are no ATs having active connections with an Access Node (AN) or Base Station (BTS), the AN still transmits the pilots. In such a scenario, each pilot burst is guarded by two 64 chip “guard”/”skirt” bursts which improves pilot acquisition and SNR computation by the ATs (in the presence of timing skew between pilot signals received from different multi-path components and different ANs). In the active case show in the diagram above, each pilot burst is guarded by two Medium Access Control (MAC) bursts.

 

Channels

The FL is logically partitioned into the following set of channels/streams:

 

  1. Pilot
  2. MAC
    1. Reverse Activity
    2. DRC Lock
    3. Reverse Power Control
  3. Traffic
  4. Control

 

Pilot

The pilot channel comprises of a simple sequence of zeros, covered by the reserved Walsh cover 64W0. This is used by all ATs in the coverage area (sector) for acquisition, synchronization, demodulation, decoding and Signal-To-Interference-And-Noise-Ration (SINR) estimation. The pilot channel is transmitted at the peak power rating configured at the AN in order to aid SINR estimation.

 

Medium Access Control

The MAC channel contains information for the set of active ATs in the sector. It also acts as a “guard” for the pilot burst in each slot.

 

The Reverse Activity sub-channel (RAC) indicates if the RL loading is too high, in which case the ATs must reduce their data rates on the RL. This is a multi-cast sub-channel intended for all active ATs in the sector. The RAB changes only every “RABLength” slots.

 

The Data Rate Control (DRC) is a mechanism introduced in EV-DO by which the AT instructs the AN as to what data rate to transmit on the FL, based on the AT's SINR measurements (more on this later in the document). The DRC Lock sub-channel indicates to the AT whether the DRC information was correctly decoded at the AN or not. A single bit is used per active AT. Please note that, the EV-DO FL is thus “rate controlled” as opposed to “power controlled” (IS-95).

 

The Reverse Power Control (RPC) sub-channel completes the control loop for regulating the power of each AT's transmission on the RL. Again, a single bit is used per active AT. The RPC bit rate is 600 bps.

 

MAC Index

The RPC and DRC Lock sub-channels are time division multiplexed on the MAC channel. The RAC is code division multiplexed with the RPC/DRC Lock. Now, whereas the RAC is multi-cast in nature, the RPC/DRC Lock sub-channels are uni-cast in nature and since the MAC channel is directed at all active ATs in the sector, the AN must “address” the RPC/DRC Lock sub-channels for the intended AT. The addressing is achieved using a “MAC Index” that is assigned to each active AT in the sector [Esteves02]. Each MAC Index maps to a 64-ary Walsh cover on the I or Q phases. Walsh covers 0-3 are reserved. 64W4 is used for the RAC. Walsh covers 5-63 can be assigned to “address” the RPC/DRC Lock sub channels. This implies a specific limit on the maximum number of ATs that maybe active at the same time, in a given sector.

 

Traffic

This stream/channel carries the user data. This stream is time division multiplexed with the control channel, on the data part of each slot. At any given instant, this stream contains information for a single AT. In order to increase the data rate to each AT, the modulation symbol sequence is de-multiplexed into 16 streams which are transmitted using a set of 16-ary Walsh covers [Bender00]. Modulation is either QPSK, 8PSK or 16-QAM.

 

Preamble

Now since the FL slots are assigned to an AT at run time (by the scheduler), the AN must provide some sort of “addressing” information so as to indicate the intended recipient AT. This is achieved using the preamble [Esteves02]. Every packet contains a preamble in the first slot (a packet maybe transmitted using more than one slot – more on this later), which contains the uncoded MAC Index. The length of the preamble is inversely proportional to the FL data rate, subject to a minimum of 64 chips for the highest data rate (of 2457.6 Kbps).

 

Control

The control channel combines the functions of the synchronization, paging and system information channels in IS-95 (meant for idle ATs in the sector). This channel is time division multiplexed with the traffic channel, on the data portion of the slot. A control channel packet is transmitted at least once every 256 slots (or 16 frames). This is known as the “control channel cycle”. There are two types of control channel “capsules” - “synchronous capsule” and “asynchronous capsule”. The “synchronous capsule” indicates the (set of) control channel packet(s) that are transmitted at time intervals that are multiples of 256 slots (the control channel packets are transmitted after a certain “offset” from the slot beginning). An “asynchronous capsule” maybe transmitted any time in-between, on a need basis. The CCH is transmitted either at 38.4 or at 76.8 Kbps.

 

Now since the control channel is time division multiplexed with the traffic channel, the AN must indicate to the AT when a data burst contains user data and when it contains control channel information. This is done by using the packet preamble which contains the uncoded MAC Index. A MAC Index of 2 indicates a 76.8 Kbps control channel packet (from this the AT can derive the number of slots needed to transmit the packet) and a MAC Index of 3 indicates a 38.4 Kbps control channel packet. Hence, these two MAC indices are unavailable for assignment to ATs.

 

Paging

The paging functionality is part of the control channel. Pages are transmitted by the AN, only within the synchronous capsule. Further, the paging cycle is fixed at 5.12 s or every 12th synchronous capsule.

 

[NSB02] describes FL channel coding/modulation, packet types, interleaver and encoder matrix designs. [Black02] deals with the FL link-budget computation. [ChungLeeMoon01] outlines FL capacity considerations. [Bi03] outlines FL data optimization techniques.

 

Reverse Link

Overview

The EV-DO RL consists of fixed size physical layer packets, each 26-2/3 ms in duration, which is equivalent to 16 consecutive FL slots. Note that the RL is not time division multiplexed like the FL and here, “slot”/”frame” is just a convenient unit of time.

 

EV-DO uses a pilot-aided coherently demodulated RL (i.e. the AT transmits a pilot on the RL in order to aid demodulation and decoding at the AN). As in IS-95, transmissions from each AT are identified by the AT's unique long PN code offset. The sectors access channel is mapped to a unique public long PN code offset, used by all ATs in that sector which wish to establish a radio link with the AN. The RL data rate may vary between 9.6 to 153.6 Kbps.

 

Channels

The RL is logically partitioned into the following set of channels:

  1. Access
    1. Pilot
    2. Data
  2. Traffic
    1. Pilot
    2. Medium Access Control
      1. Reverse Rate Indicator
      2. Data Rate Control
    3. ACK
    4. Data

 

Access

The access channel is used by the AT to initiate communication with the AN or to respond to a page. The AT's first transmits an “access probes” and waits for a response from the AN. If there is no response, the AT transmits another access probe, at a higher power level. Each access attempt thus consists of a series of such access probes. If one access attempt fails, the AT waits for a certain period of time before initiating a new access attempt. The wait period between successive access probes in the same access attempt, the power step increase for successive access probes, etc. are all system parameters that are broadcast by the AN on the control channel. The situation is depicted pictorially below:

 

 

Each access probe consists of a preamble followed by an access channel data packet. The preamble duration is 1 frame, or 16 slots. During the preamble, only the pilot channel is transmitted. The access channel datapacket is transmitted over 4 frames. During the access channel data packet, both the pilot and access data packets are transmitted using code division multiplexing.

 

Traffic

The RL traffic channel structure at the physical layer is as follows:

As we see from the above two diagrams, the RL traffic channel consists of four code division multiplexed information streams, two streams per I and Q phases. As already mentioned, the user data is a fixed size 16 slot-duration packet, on the Q phase. Also on the Q phase is the DRC information stream (more on this later). On the I phase, the ACK information stream is transmitted on each slot, for half a slot. The RRI and pilot streams are time division multiplexed onto the I phase: the first 256 chips of each slot contain the RRI information and the rest of the chips are dedicated to the pilot.

 

For identification, each information stream is covered by a (known to the AN) Walsh cover. Since the Walsh covers are only used to identify the information streams within an AT's traffic channel, 16-ary and 8-ary Walsh covers are used instead of 64-ary ones. The pilot/RRI information stream is covered by 16W0. The ACK information stream is covered by 8W4. The DRC information stream is covered by 16W8 and the user data stream is covered by 4W2.

 

Reverse Rate Indicator

The RL MAC channel contains two streams of information: the RRI and the DRC. The RRI indicates to the AN the rate at which the AT is transmitting on the RL. The RRI is included as the preamble for RL frames. Essentially this boosts the RL capacity as it aids coherent decoding and demodulation of RL frames at the AN. An 8-ary orthogonal code is used to indicate the rate.

 

Data Rate Control

The DRC MAC sub-channel is used by the AT to indicate to the AN the desired rate of transmit on the FL as well as the specific AN sector on which to transmit [Cui02]. The data rate is requested using 8-ary bi-orthogonal coding. The desired AN sector is requested using an 8-ary Walsh cover. Each DRC slot contains 1024 chips and the data is centered around the middle of the slot so as to minimize the delay between SINR estimation and the start of AN transmission.

 

ACK

The ACK information stream is related to the use of HARQ on the EV-DO FL in order to improve system throughput under changing channel conditions (more on this later). It indicates whether the last transmitted packet was received and decoded correctly at the AT or not.

 

Adaptive Modulation

The cellular wireless environment changes constantly. Terminals in a cellular network experience fading as a function of time, frequency, location, speed, etc. In order to maximize the FL throughput, the EV-DO air interface protocol tracks these changes in the air-link conditions and adapts the code rate, data rate and modulation to the changing channel SINR.

 

The AT uses the 2 burst pilots transmitted per slot on the FL to estimate the channel SINR (averaged over a number of slots, = “DRCLength”). Based on the measured SINR, the AT derives the maximum transmit rate on the FL at which a target BER/FER threshold is maintained and indicates this to the AN using the RL MAC DRC sub-channel.

 

The following set of FL transmit rates are used:

 

Data Rate (Kbps) Slots per Packet Packet Size (bits) Code Rate Modulation Preamble (chips) Effective code rate
38.4 16 1024 1/5 QPSK 1024 1/48
76.8 8 1024 1/5 QPSK 512 1/24
153.6 4 1024 1/5 QPSK 256 1/12
307.2 2 1024 1/5 QPSK 128 1/6
307.2 4 1024 1/3 QPSK 128 8/49
614.4 1 1024 1/3 QPSK 64 1/3
614.4 2 2048 1/3 QPSK 64 16/49
921.6 2 3076 1/3 8PSK 64 16/49
1228.8 1 2048 1/3 QPSK 64 2/3
1228.8 2 4096 1/3 16QAM 64 16/49
1843.2 1 3076 1/3 8PSK 64 2/3
2457.6 1 4096 1/3 16QAM 64 2/3

 

Table 1: FL Transmit Rates

 

As we see from the table, at lower rates, multiple slots are required to transmit a single packet. The length of the preamble (in terms of chips) decreases with high transmit rates, subject to a minimum of 64 chips.

 

Further, the AT typically receives signals from multiple sectors, on the FL. In this scenario, the SINR for user k is:

SINR(k) = Σ SINRj(k) βj(k) : j = all sectors in active set j

where, SINRj(k) = SINR from sector j seen by user k and, βj(k) = fraction of power from sector j allocated to user k

 

The optimum normalized power allocation is achieved when the total power is allocated to the best link, maximizing SINR and thus the data rate. Hence, the AT also measures the SINR of all the pilots in its active set and determines the pilot with the highest SINR. It then uses the RL MAC DRC sub-channel to instruct the AT to transmit data to the AT only from this sector. The scenario is depicted pictorially below:

 

 

Hence, on the FL, there is no concept of soft hand-off (with the resulting over heads) which increases the FL capacity. As shown in the diagram, there is a short packet transmit delay while the new sector indicated by the AT is decoded by the AN and the queued packets at the previous sector (“A” in the diagram) are re-sent to the new sector (“B” in the diagram). Also, although not explicitly shown in the diagram, as the SINR for the sector changes, the DRC also indicates different FL transmit rates. The above mechanism of changing the serving AN sector dynamically is referred to as “fast cell site selection” or “adaptive FL server selection” or “virtual soft hand-off”.

 

[Yavuz03] describes the link adaptation mechanism. [Leela04] and [Huang02] outline throughput performance of the FL and RL.

 

Hybrid ARQ

The above mechanism of adapting to the Channel State Information (CSI) of the wireless link is limited by the fact that, effects due to fast fading and intersector interference cannot be predicted and hence the accuracy of the a-priori CSI varies considerably with channel conditions and interference patterns. As a result, the data rates indicated to the AN by the AT, based on the a-priori CSI are quite conservative in nature. Hence, in order to maximize the spectral efficiency and the FL throughput, a-posteriori CSI feedback is required. In EV-DO, this is achieved by a two-pronged approach:

 

1. Usage of turbo coding techniques.

2. Usage of hybrid_ARQ.

 

As we have seen in the previous section, at lower transmit rates (worsening channel conditions), a single packet transmission on the FL requires multiple slots. However, due to the usage of turbo codes, it is possible for the AT to decode and recover the full packet data even from the first slot transmission. If the AT is unable to derive the packet data from the first slot, the data in the second transmitted is re-combined with the data already received in the first slot such that, the probability of successfully deriving the packet data is now greater. And this is true with each subsequent slot transmission. Until we reach the maximum number of slots for the given transmit rate. The above mechanism is known as “incremental redundancy”. Note, this also implies that HARQ does not provide any advantages at the highest transmission rates where a single slot is used to transmit the packet data.

 

In order to take advantage of incremental redundancy, it is thus necessary to insert a “wait period” between the transmission of successive slots and to wait for feedback from the AT as to whether the packet data has been successfully decoded or not. Hence EV-DO uses a “4-slot interlacing” scheme as follows:

 

 

In the diagram above, the notation “1A” for the FL traffic channel stands for “the first slot (of a multi-slot transmission) of packet 1”. The following points are important:

  1. The diagram shows that the AT has requested for a FL transmission rate of 153.6 Kbps using the RL DRC MAC sub-channel. This implies that the packet data is transmitted in a maximum of 4 slots.
  2. So the AN transmits the first of these slots.
  3. In the next 3 slots, the first slot of the next 3 data packets are transmitted. These data packets maybe destined to the same AT or to different ATs.
  4. Meanwhile, on the RL ACK MAC sub-channel, the AT transmits a “NAK” which signifies that it has not been able to derive the packet data from the first slot transmission.
  5. Based on the “NAK”, after transmitting the first slots of packets 1-4, the AN transmits the second slot of packet 1.
  6. In the next FL traffic channel slot, we see “2B” which implies that the second slot for packet 2 has been transmitted by the AN. That also means that the FL rate for packet 2 is such that it is a multi-slot packet and that the ACK channel response to the first slot for packet 2 is also a “NAK” (not shown).
  7. This is followed by the first slots for packets 5 and 6.
  8. We again see that the AT has responded with a “NAK” for the second slot for packet 1 and so the AN transmits the third slot.
  9. On receiving the third slot for packet 3, we see that the AT transmits an “ACK” on the ACK channel. This means that the AT has been able to derive the full packet data. Also, this means that the transmission did not require the full 4 slots. This is known as “early termination” and implies that the actual data rate is greater than 153.6 Kbps.
  10. On receiving the “ACK”, the AN simply never transmits the fourth slot for packet 1 and instead proceeds to transmit the first slot for packet 9.
  11. In the diagram, please note that the implicit timing relation between the DRC, ACK and FL traffic channels is an important component in the HARQ mechanism and simplifies the feedback channel design.

 

In summary, HARQ uses the excess Eb/N0 (due to conservative maximum data rate estimation based on a-priori CSI) to terminate packet transmissions early (wherever possible) leading to a higher effective throughput. For more on HARQ, please refer to [Das01], [Falahati] and [Yavuz01].

 

RL Power_Control & System Stability

As described above, the EV-DO RL is code division multiplexed as in IS-95 and hence it is interference limited. Thus, as in IS-95, power control becomes important in order to achieve an acceptable FER with minimum AT transmit power (thus maximizing sector throughput as a whole). EV-DO uses the same set of three power control loops as in IS-95:

 

  1. Open Loop control: This is implemented by the AT based on its measurement of the FL pilot strength – the stronger the measured pilot strength, the closer the AT is to the AN (the pilot is transmitted at full AN power) and hence the lower should be the RL transmit power.
  2. Inner Closed Loop control: The AN sends commands 800 times a second to increase or decrease the AT's transmit power on the RL. The network sets an “operating point” (also known as the Power Control Threshold) for each AT, based on the System operating point (more on this later). This power control loop attempts to hold the AT's RL transmit power to the set PCT.
  3. Outer Closed Loop control: The AN modifies the AT's PCT from time to time in order to achieve a given FER target. The power control loops described above, determine the transmit power only for the RL pilot channel. The strengths of all the other channels is determined relative to the pilot strength.

 

In addition to RL power control, it is important for the AN to also control the RL data rates for each AT and the overall sector load, in order to ensure a stable system operation. The system stability depends on the “operating point” which is defined by the metric “Rise over Thermal” (RoT):

 

RoT = Total received power (I0) – Thermal Noise floor (N0)

 

In order to ensure stability, the sector RoT must be maintained below a target threshold (the threshold is determined by the link budget and the desired margin of stability). This is done by setting the RAB on the RAC, every frame, on the FL (RAB = 1 if the measured RoT is greater than the desired system operating point). Further, the maximum RL transmit rate for that sector is broadcast on the control channel. The AT processes the RAB as follows:

  1. Perform a logical OR operation for all RAB information received for all pilots in its “active set”.
  2. If the cumulative RAB = 1 (busy), reduce to the next lower transmit rate with probability “q”.
  3. If the cumulative RAB = 0 (not busy), increase to the next higher transmit rate with probability “p”
  4. Probabilities “p” and “q” depend on the current transmit rate and are defined as part of the protocol standard.
  5. The AT maintains the current transmit rate with probability (1 – p) or (1 – q) respectively.
  6. The AT RL transmit rate however need not go below 9.6 Kbps.
  7. The AT transmit rate has to be less than or equal to the maximum RL transmit rate advertised for that sector.

Now, in order to accurately measure the RoT metric, EV-DO introduced a “silence period” for the duration of which, all active ATs in the sector cease transmission. The silence interval is usually between 1-3 frames and the period can be 54 s or 109 s and so on. All parameters pertaining to the silence period are broadcast by the AN over the control channel.

 

Finally, combining RL power and rate control, based on the available power budget and the RL transmit rate derived from the RAB, the AT boosts the transmit power for the other channels on the RL, relative to the pilot power (which is directly derived from the power control mechanism). For example, at a target data rate of 9.6 Kbps, the data channel power is boosted with respect to the pilot power by (say) 4 dB. The actual power at which transmission occurs will depend on the pilot transmit power, which is controlled by the power control loops. Thus we see that, the RL transmit power and data rates are closely inter-linked.

 

[Yeo03], [Lee05] and [Yavuz02] propose modified RL rate control algorithms. [YeoCho] presents a Markov model of the RL rate control scheme.

 

FL Scheduling

As has already been discussed before, the FL transmissions are time division multiplexed. Further, each time slot is not preassigned to an AT but changes dynamically. The situation is further complicated by the HARQ requirement of multi-slot transmissions of data packets, inter-laced with other packets. So, the heart of the EV-DO FL is the FL scheduler. The scheduler needs to fulfill two objectives:

  1. Maximize the sector throughput.
  2. Ensure short-term and long-term fairness in terms of each active AT being allocated at least some time slots on the FL.

To meet the first objective, the scheduler takes into account the a-priori CSI provided by the DRC channel.

 

The simplest scheduling approach ([Chang02] and [Elliot02]) would be to transmit to the AT with the highest average channel SINR at any instant. The metric is:

 

mi(t) = DRCi(t)/Ravgi(t)

where DRCi(t): Instantaneous requested rate of user “i” at time “t” and Ravgi(t): Average throughput of user “i” at time “t”.

 

The average throughput Ravgi(t) is computed as follows:

 

Ri(t + 1) = (1 – γ)· Ravgi(t) + γ Ri(t)

γ = 1/tc where tc is the scheduler time constant in slots and Ri(t) is equal to the requested rate of user “i” in the slot at time “t” if the user was selected for transmission in that slot or 0 otherwise. “tc” represents the maximum time an AT may have to wait for allocation of a time slot.

 

This is known as the “proportional fairness scheduler” since the scheduling quantum is proportional to the ATs average SINR. This approach obviously maximizes the sector throughput by taking advantage of what is known as the “multi-user diversity gain”, which increases with the number of active ATs in the sector as well as increased fluctuations in individual channel conditions. It also meets the requirements of long-term fairness since each MT would sooner or later experience good channel conditions. However, in the short term, this approach can be patently unfair with large variations in the highest and lowest user transmission rates.

 

An alternate scheduling approach [Jalali00] which attempts to overcome the short-term fairness shortcoming above is known as the "generalized proportional fairness scheduler. The modified metric is:

 

mi(t) = DRCi(t)/Davgi(t) . h(Davgi(t))/ Ravgi(t)

where Ravgi(t) is the average throughput as before and Davgi(t), the average DRC is computed as:

 

Davgi(t + 1) = (1 - γ)·Davgi(t) + γ DRCi(t)

h(x) specifies the “fairness criteria” which maybe adjusted as desired. γ is the same as before. The function h(x) maybe used to define the user defined fairness criteria while the first term controls the multi-user diversity.

 

The FL scheduler has not been rigorously defined in the EV-DO standards as this has been an area of active research. See [FanHuang04], [ChoiHan02], [Ban04], [Kim02] and [Huang03] for descriptions of alternative approachs. See Generalized_Processor_Scheduling for more discussion.

 

EVDO Signaling Protocol Stack

The EV-DO protocol stack [Bender00] enables a modular design that allows partial updates to protocols and software and independent protocol negotiation. The layers of the protocol stack are:

 

  1. Physical: This layer deals with channel structure, frequency, power output, modulation and encoding specifications for the FL and RL.
  2. MAC: This layer comprises of 4 protocols:
    1. Control Channel MAC: This deals with the control channel that broadcasts all important information controlling AN transmission, packet scheduling, AT acquisition and AT packet reception.
    2. Access Channel MAC: This deals with the timing, power and other protocol requirements for sending messages on the access channel.
    3. Forward Traffic Channel MAC: This deals with sending user data packets at optimal packing efficiency by utilizing variable and fixed transmission rates and ARQ interlacing. This also deals with rules to interpret the DRC channel on the RL as well as rules for DRC supervision.
    4. Reverse Traffic Channel MAC: This deals with receiving the user data packets on the RL by acquiring the RL and interpreting the RL data rate selection.
  3. Security: This deals with ensuring the security of the air interface connection between the AT and the AN. It provides the functions of key exchange (using the Diffie-Hellman key exchange protocol to exchange keys necessary for authentication and encryption), authentication and encryption.
  4. Connection: This layer comprises of a group of protocols that combine to efficiently manage the air link, reserve resources, etc:
    1. Air Link Management: Manages the state of the air link between the AT and the AN by triggering other protocols based on the state.
    2. Initialization State: Deals with actions associated with acquiring a EV-DO network such as network determination, pilot acquisition and system synchronization.
    3. Idle State: Deals with the scenario wherein the AT has acquired the network but is neither sending nor receiving data. This protocol provides procedures for opening a connection, conservation of AT power and triggering the Route Update protocol when required.
    4. Connected State: Deals with the scenario wherein the AT has a connection with the network and is either sending or receiving data or both. This protocol provides procedures for hand-offs, connection closure and so on.
    5. Route Update: This protocol constantly reports the current sector and the set of potential neighboring sectors to the AN, to support AT mobility via hand-offs.
    6. Overhead Messages: Broadcasts essential parameters pertaining to the operation of other protocols over the control channel.
    7. Packet Consolidation: Deals with ensuring negotiated QoS is met for user data.
  5. Session: This layer comprises of a set of three protocols that provide a support structure for the lower layers in the protocol stack:
    1. Session Management: This acts as an overall controller as well as handles session lifetime management and cleanup.
    2. Address Management: This deals with maintaining the AT address and assignment of a locally unique ID to the AT.
    3. Session Configuration: This deals with procedures to negotiate and provision the set of protocols used during the session as well as the configuration parameters for these protocols.
    4. Stream: This is a multiplexing/de-multiplexing layer that works along with the Packet Consolidation protocol to prioritize user data and the signaling traffic appropriately.
  6. Application: This comprises of a suite of protocols that increase the robustness of the underlying layers of the stack. The layer is partitioned into two sub-layers based on signaling and user data traffic:
    1. Default Signaling Application:
      1. Signaling Network Protocol (SNP): Provides a message transmission service for signaling messages originating fromother protocols.
      2. Signaling Link Protocol (SLP): Provides the transport for the SNP messages offering fragmentation and re-assembly as well as best effort and reliable delivery options.
    2. Default Packet Application:
      1. Radio Link Protocol (RLP): Uses a NAK-based re-transmission scheme to achieve lower FER than provided by the lower layers of the protocol stack.
      2. Location Update Protocol (LUP): Provides mobility management using the Route Update protocol in the lower layers.

 

 

Standards and References

 

[TIA-856-1]: cdma2000 High Rate Packet Data Air Interface Specification

 

[TIA-878]: Inter-Operability Specification (IOS) for High Rate Packet Data (HRPD) Access Network Interfaces

 

[TIA-864]: Recommended Minimum Performance Standards for cdma2000 High Rate Packet Data Access Network

 

[TIA-866]: Recommended Minimum Performance Standards for cdma2000 High Rate Packet Data Access Terminal

 

[TIA-890]: Test Application Specification (TAS) for High Rate Packet Data Air Interface

 

[TIA-919]: Signaling Conformance Standard for cdma2000 High Rate Packet Data Air Interface

 

[TIA-925]: Enhanced Subscriber Privacy for cdma2000 High Rate Packet Data

 

[TIA-835]: CDMA2000 Wireless Internet Protocol (IP) Network Standard

 

[TIA-2001]: Inter-operability Test Specification for CDMA2000 Access Network Interfaces

 

[Bender00] Paul Bender et. al.; “CDMA/HDR: A Bandwidth-Efficient High-Speed Wireless Data Service for Nomadic Users”, IEEE Communications Magazine, pp 70-78, vol 38, July 2000

 

[NSB02] Nagabhushana T. Sindhushayana and Peter J. Black; “Forward Link Coding and Modulation for CDMA2000 1xEV-DO (IS-856)”, PIMRC 2002, Lisbon, Portugal, September 2002

 

[Black02] Peter J. Black and Qiang Wu; “Link Budget of cdma2000 1xEV-DO Wireless Internet Access System”, PIMRC 2002, Lisbon, Portugal, September 2002

 

[Paran01] David W. Paranchych; “1xEV-DO Network Design and Performance”, pp 153-156, 2001 IEEE

 

[Leela04] Rangsan Leelahakriengkrai et al.; “Performance Analysis of 1xEV-DO Systems under Realistic Traffic Models and Limited-Size IP Backhaul”, 10th Asia-Pacific Conference on Communications and 5th International Symposium on Multi-Dimensional Mobile Communications, 2004

 

[ChungLeeMoon01] Wonsuk Chung, Hong Woo Lee and Jungbae Moon; “Downlink Capacity of CDMA/HDR”, pp 1937-1941, VTC '01, 2001 IEEE

 

[Cui02] Dongzhe Cui et al.; “Reverse DRC Channel Performance Analysis for 1xEV-DO: Third Generation High-Speed Wireless Data Systems”, pp 137-140, 2002, IEEE

 

[Bi03] Qi Bi et al.; “Performance of 1xEV-DO Third-Generation Wireless High-Speed Data Systems”, Bell Labs Technical Journal, 7(3), pp 97-107, 2003

 

[Yavuz03] Mehmet Yavuz and David W. Paranchych; “Adaptive Rate Control in High Data Rate Wireless Networks”, pp 866-871, 2003 IEEE

 

[Yeo03] W. Y. Yeo and D. H. Cho; “Enhanced Rate Control Scheme for 1xEVDO reverse traffic channels”, Electronic Letters, No 23, Vol 39, 13 th November, 2003

 

[Esteves02] Eduardo Esteves, Peter J. Black and Mehmet I. Gurelli; “Link Adaptation Techniques for High-Speed Packet Data in Third Generation Cellular Systems”, Euro. Wireless Conf. 2002

 

[Huang02] Ching Yao Huang et al.; “Forward and Reverse Link Capacity for 1xEVDO: Third Generation Wireless High-Speed Data Systems”, pp 871-875, 2002 IEEE

 

[YeoCho] Woon-Young Yeo and Dong-Ho Cho; “A Markovian approach for modeling IS-856 reverse link rate control”, pp 3236-3240, IEEE Communications Society

 

[Lee05] Hye Jeong Lee et al.; “New Rate Control Scheme Based on Adaptive Rate Limit for 1xEV-DO Reverse Link Traffic Channels”, IEEE Communication Letters, Vol 9, No 10, October 2005

 

[Yavuz02] Mehmet Yavuz and David W. Paranchych; “A Method for Outer Loop Rate Control in High Data Rate Wireless Networks”, pp 1701-1705, 2002 IEEE

 

[FanHuang04] Yu Long Fan and Ching Yao Huang; “Adaptive Scheduler Algorithm for Multi-Criterion Designs in B3G Wireless Systems”, WCNC 2004, pp 1293-1297, IEEE Communications Society

 

[ChoiHan02] Yoonghoon Choi and Youngnam Han; “A Channel-based Scheduling Algorithm for cdma2000 1xEV-DO System”, PIMRC 2002, Lisbon, Portugal, 2002

 

[Ban04] Taewon Ban; “A New Efficient Scheduling Algorithm for the support of QoS in cdma2000 1xEV-DO”, pp 1148-1152, 2004 IEEE

 

[Kim02] Kuenyoung Kim et al.; “A Proportionally Fair Scheduling Algorithm with QoS and Priority in 1xEV-DO”, PIMRC 2002, Lisbon, Portugal, 2002

 

[Chang02] Kapseok Chang and Youngnam Han; “QoS-based Adaptive Scheduling for a Mixed Service in HDR System”, PIMRC 2002, Lisbon, Portugal, 2002

 

[Huang03] Ching Yao Huang et al. and Stan Vitebsky et al.; “Schedulers for 1xEVDO: Third Generation Wireless High-Speed Data Systems”, pp 1710-1714, 2003, IEEE

 

[Elliot02] Robert C. Elliot and Witold A. Krzymien; “Scheduling Algorithms for the cdma2000 Packet Data Evolution”, pp 304-310, 2002, IEEE

 

[Jalali00] A. Jalali, R. Padovani and R. Pankaj; “Data Throughput of CDMA/HDR a high efficiency high data rate personal communication system”, Proc. IEEE 51st VTC, Tokyo, May 2000

 

[Das01] Arnab Das, Farooq Khan and Sanjiv Nanda; “A2IR: An Asynchronous and Adaptive Hybrid ARQ Scheme for 3G Evolution”, pp 628-632, VTC'01, 2001 IEEE

 

[Falahati] Sorour Falahati and Arne Svensson; “Hybrid Type-II ARQ Schemes with Adaptive Modulation Systems for Wireless Channels”

 

[Yavuz01] Mehmet Yavuz, David Paranchych and Geng Wu; “Performance Improvement of HDR System due to Hybrid ARQ”, pp 2192-2193, 2001, IEEE

 

[Semyon07] Semyon Mizikovsky, Zhibi Wang and Hongru Zhu; “CDMA 1x EV-DO Security”, Bell Labs Technical Journal, 11(4), pp 291-305, 2007

 

 

 


Page Information

  • Changed 6 months ago [show history]
  • View page source
  • You're not logged in
  • Spam-like content was removed from this page.
  • No tags yet learn more

Wiki Information

Recent PBwiki Blog Posts