Handoff Details

Good handoff performance in CDMA is not just a nicety; it is absolutely essential. A mobile that is being served by one base station when another is closer in terms of path loss will be transmitting more power than would be necessary were it using the "right" cell. The fact that that mobile, and all others like it radiate excess power raises the overall interference level. The higher overall interference level increases the effective reuse factor, and thus reduces overall, average reverse link capacity. Sloppy, late, or slow handoffs thus should be kept to an absolute minimum.

On the other hand, there is a detrimental effect of handoff due to the asymmetry in power control design between forward and reverse links. While the reverse link power control is fast and accurate, the forward link power control is slow and loose. In IS-95A it is implemented via messaging. It is somewhat better in J-STD-008, where there is an erasure report bit in each reverse frame; this permits a faster power control implementation.

Soft handoff requires that multiple base stations transmit the same traffic to the mobile in question. The multiple active forward channels raise the overall interference level at the mobile - like the reverse link, this increases the effective frequency reuse factor and thus reduces forward link capacity.

But perhaps we are a little ahead of ourselves. First, let us define what is meant by soft handoff in CDMA.

What happens during soft handoff?

CDMA soft handoff is a call state in which two or more base stations support a mobile station. Those stations can be either separate sectors of separate cells, or they can be multiple sectors of the same cell, or any combination of these. The inter-sector handoff, in fact, is very common because of the broad antenna gain patterns and the vagaries of urban multipath propagation.

Forward CDMA Channel

Each participating cell in a soft handoff transmits the same traffic stream to the mobile, bit-for-bit. They do so on any available code channel. Each base station chooses a code channel simply on the basis of availability. The mobile station must implement, in its Rake receiver, multiple fingers that are capable of "tuning" to any of the 63 available code channels. The outputs of those Rake fingers must be combined for good E_b/N₀ performance. The presence of a pilot in the Forward CDMA Channel allows optimum coherent combining of those Rake outputs.

Reverse Power Control

Embedded in the Forward CDMA Channel are the reverse power control bits. These occur in pseudo-random positions in each 1.25 ms interval (power control group), or 16 times per frame. Each power control bit is interpreted as a command to raise or lower power by an increment of approximately 3/4 dB.

Each base station makes power control decisions independently. The mobile station is responsible for demodulating the power control bits and raising or lowering its power accordingly. The goal of the power control is to maintain the reverse link transmit power at the lowest possible level commensurate with adequate error performance. The mobile is thus required to interpret the power control bits, which will often disagree, as requiring an increase in power only if all base stations in the handoff say "up"; if any participant says "down" then the mobile is required to reduce power. This rule is sometimes called "OR of the downs" - if anybody says down, you go down.

Reverse CDMA Channel

Spreading of the Reverse CDMA Traffic Channel is mobile-unique. There is nothing about the coding and modulation that depends in any way on the base stations that are serving the mobile. The mobile thus needs do nothing special about handoff, aside from proper interpretation of the power control bits.

Combining of the reverse link signals to the base station is not specified in the air interface or performance standards. However practical considerations strongly encourage the use of selection diversity. That is, each base station demodulates, deinterleaves, and decodes the traffic independently. When the traffic frames from the handoff participants arrive at the network interface, frame quality metrics can be compared and the best frame chosen for transmission to the network. This is sometimes called selection diversity - Use the best of the N available copies of each traffic frame.

What is "softer" handoff?

All of this is modified a bit if the participants in a soft handoff are sectors of the same cell. This situation has come to be known as "Softer" handoff. Collocated stations permit combining to be done in a CDMA modem that has visibility of multiple sectors. Such combining can be done on a symbol-by-symbol basis, rather than by selection of entire frames.

Likewise, a common modem that services all sectors of a handoff is capable of sending identical power control bits on all the sectors. This is allowed for in the air interfaces. The handoff direction message that initiates handoffs contains a field that shows which stations are transmitting the same power control bits so that the mobile can do pre-decision combining of those bits.

To the mobile, a softer handoff is identical to soft handoff except for the treatment of the power control bits.

What initiates soft handoff?

Pilot Search

CDMA is said to use Mobile Assisted Handoff (MAHO). In practice this means that the mobile station continuously searches for the pilot code using a PN correlator specifically designated for this purpose. Universality of the pilot code (or Short Code) facilitates the search. All base stations use the same code. The mobile station can search in timing hypothesis without having to change the PN sequence.

If the mobile already has a notion of CDMA system time, as it does if it is already involved in a call, then it can report the relative timing of a newly detected pilot. What distinguishes base stations from one another is the phase of their pilots. The period of each pilot is 26.667 ms. They are separated by a minimum of 64 chips, which is about 52 ms or about 15 km at the speed of light. The mobile timing will normally be good enough that a reported pilot offset unambiguously identifies the base station it has detected.

Detection Thresholds

The mobile reports pilots on the basis of their pilot-to-interference ratio (PIR). The PIR (called, strangely, E_c/I₀ in much of the literature and standards) is compared to an absolute threshold to determine when it should be reported as a handoff candidate. That threshold is a parameter that the mobile obtains from the overhead messages broadcast by the base stations. When a pilot crosses the first threshold, T_ADD, then its presence is reported, via a message, to the network. The network will normally add that base station to the so-called Active Set, that is, the set of base stations that are participants in the soft handoff to the mobile in question.

The second threshold is not absolute but relative. It is compared to the difference between the largest PIR in the active set and the PIRs of all other members. When any of them falls below this threshold, T_DROP, then another message is transmitted. The normal result is that the base station in question will be dropped from the Active Set, and that will be reported to the mobile by a signaling message.

The effect of the two thresholds, one absolute, the other relative, is to ensure that any station that is able to contribute in any significant way to the overall SNR after diversity combining, is in the active set with high probability. Conversely, a base station is dropped only when it has deteriorated far below the best station. If the best station is itself marginal then the next strongest station will be retained. This two-threshold scheme has been found in practice to be very effective. Its only possible drawback is that it sometimes errs on the side of too much handoff. Too much handoff reduces capacity because of the excess number of Forward Traffic Channels needed to support it. It also impacts the number of channel elements (CDMA modems) needed in the base stations.

Does CDMA always use soft handoff?

No. It does so whenever possible because the performance is very much superior to other forms of handoff. However there are several forms of handoff that cannot be done "softly".

Interfrequency Handoffs

If there are multiple CDMA carrier frequencies active, then handoffs between them must be hard. While inter-frequency handoff is physically possible, the decision was made in the standards committees to not require it. To implement it would very much complicate the mobile station. Multiple (well at least two) independent frequency synthesizers would be required. Pilot searches would have to span all active frequencies. All of this would very much increase the cost of the subscriber sets, this in a very much cost-driven marketplace. While soft inter-frequency handoff would certainly improve performance, the marginal benefit of it was deemed to not justify the cost.

Inter-frequency hard handoffs are probably best accomplished by doing so within one geographical site, rather than trying to hand off to a neighboring site. The inter-frequency handoff can first be executed intra-site, where the mobile's timing is already known, followed immediately by a soft handoff to the neighbor without frequency change on the new frequency.

Timing Changes

The CDMA air interfaces, to allow load balancing on network transmission facilities, include the ability to offset traffic frame timing from system time. Those timing offsets are accomplished by a hard handoff.

Digital-to-Analog

Quite obviously there is no practical way to do soft handoffs between a CDMA digital system and an analog FM system. These handoffs are thus inherently hard.

What about handoffs between FM and CDMA systems?

CDMA to analog handoffs are carried out by sending the same information that is required in a normal analog to analog handoff. Analog to digital handoffs are not allowed.

Again, A-to-D handoff is not physically impossible, but to do so would require retro-fit of all the analog sites that might need to do so. Moreover, it was concluded in the standards committees that the real operational need for A-to-D handoff was dubious at best. Analog base stations are not going to disappear at the first hint of CDMA service. The CDMA system is specifically designed to permit graceful changeover from all-analog to primarily digital service over a span of many years. The CDMA service will thus be introduced as an overlay to the existing analog infrastructure. A mobile that is involved in an analog call and moves into a CDMA service area can continue to receive service on the analog system. At worst, if an analog-to-analog handoff fails at the boundary, the subscriber will have to re-originate the call within the CDMA service area. At that time the dual mode handset will execute a system determination (rescan, in the old analog terminology), and will discover that CDMA service is available. Assuming the subscriber has not specifically de-selected CDMA service, the unit will do the re-origination on the CDMA system. While this is not the most elegant solution, it is not greatly different than the service the analog customers have been receiving for several years.

The complexity and cost impact of trying to introduce a viable A-to-D handoff was thus deemed unjustified.

In this connection, there is the question of how do CDMA mobiles discover and identify AMPS handoff candidate cells. There are several possible solutions to this that we will discuss at a future time.

| Index | Topics | Glossary | Standards | Bibliography | Feedback |

w

Copyright © 1996-1999 Arthur H. M. Ross, Ph.D., Limited