By Ramzi Bargouthi

The coming of IS-95 networks promised much in terms of new capabilities, but with a new technology came new approaches and new tasks for cellular engineers to learn. What are the major challenges of cdmaOne network planning and optimization? Ramzi Bargouthi, CDMA technical director, MSI

Code Division Multiple Access (CDMA) technology has come a long way since the late 1980s when it was first proposed by Qualcomm Inc. as an alternative to TDMA and FDMA techniques to enhance the capacity of cellular networks. Initially, CDMA was regarded with skepticism, but it was a skepticism mingled with excitement-an excitement that derived from two main factors.

First there was the promise of high capacity, and hence a reduction in the number of base stations, compared to the existing analog or proposed TDMA systems. The second factor was the inherent CDMA capabilities that on one hand make the design and optimization of the network seem less involved than that of other systems, and on the other make the real-world operation of the network less complicated. These capabilities are:

o frequency reuse: all users on a CDMA carrier transmit over the same RF channel, eliminating the challenging tasks associated with frequency planning;
o cell breathing: as the traffic loading in a CDMA cell increases, the cell will automatically shrink in size while surrounding cells expand to compensate for the resulting drop in the coverage area;
o soft capacity limit: as a cell shrinks to its minimum size allowed by the infrastructure, it can still accommodate more users with a graceful degradation in call quality; and
o soft handoff: a user may enter handoff with another cell and continue to be connected to the original cell for some time before dropping it. This reduces dropped calls due to handoff failures, eliminates 'ping-pong' effects and also improves system capacity and call quality while in soft handoff when the links act as a form of diversity.

Today numerous IS-95 based CDMA networks, now known as cdmaOne systems, are operational worldwide, and most of these expectations have proved true. However, the new systems brought with them a whole new set of challenges for cellular engineers involved in designing, and optimizing the performance of, cdmaOne networks.

The major tasks are:
o maximizing system capacity;
o maximizing system performance as it applies to improved call quality, sufficient cell coverage and minimum dropped calls; and
o minimizing cost (both deployment and operational).

Below, we discuss some of the principles and challenges associated with the design and optimization of cdmaOne cellular networks. It is assumed that the reader is familiar with the basic principles of these systems.

Frequency reuse efficiency and cdmaOne capacity
cdmaOne systems employ universal frequency reuse where all users on a cdmaOne carrier transmit over the same RF channel. Along with this simplicity of frequency planning come the challenges of designing the network and setting the system parameters to optimize the frequency reuse efficiency. Simply, the capacity of a cdmaOne cell drops as adjacent cdmaOne cells are deployed. The reason for this is that the users of these cells are sharing the same RF channel and thus increasing the noise floor seen at the receiver. In the absence of signals from other cells, the noise floor would only result from users within the cell; in this case, the cell capacity is at a maximum. This represents a cell with a frequency reuse efficiency of 100 per cent. As signals from other cells share the RF channel, the frequency reuse efficiency drops in accordance with the amount of interference they contribute as compared to the interference from within the cell. This is described in more detail below.

For a single, isolated cdmaOne cell, the noise floor seen at the receiver is the summation of the background noise, control channel power and traffic channel power of users within the cell. The cell capacity in this case is limited (soft limit) by the noise floor and can be determined by solving for N users in the equation below. This equation applies to the cdmaOne reverse link with power control. A similar equation can be derived for the forward link:
Desired SNR = S / [(N-1) v S + No] = S / [Isc + No]

Where:
Isc = Interference power from users within the cell
S = Received signal power, assumed equal for all users due to power control
v = Voice activity factor
No = Background noise power

Now, for a multi-cell network, the noise floor is increased by the interference received from users in other cells, Ioc, and the capacity equation is:
Desired SNR = S / (Isc + Ioc + No) and the frequency reuse efficiency is, by definition:
f = Isc / (Isc + Ioc)

Notice here that the SNR, and cdmaOne system capacity, drop as a function of Ioc; the challenge is thus to design cell site locations, antenna orientations and system parameters to minimize Ioc. This does not mean, however, that Ioc can be minimized to zero by isolating the cdmaOne cell. The reason for this is that in cdmaOne some overlap between cells is needed to allow soft handoff and hence a minimum value of Ioc would always be present at the receiver. Therefore a compromise between cell coverage and soft handoff parameter settings needs to be achieved to minimize Ioc.

It is also noteworthy, looking at the second equation above, that the desired SNR and system capacity improve with a drop in the total interference seen at the receiver. That brings us to one of the major inherent capabilities of cdmaOne: a technique that results in a reduction in system interference translates directly into an improvement in system performance and capacity. This is the principal behind the improvements in cdmaOne systems performance when deploying interference cancellation techniques, antenna sectorization, smart antennas, power control and more.

Soft handoff and handoff parameters optimization
In a cdmaOne network, a mobile station (MS) travels between the coverage areas of multiple cells and constantly monitors the strengths of the pilot channels. Following IS-95 protocols the MS is expected to enter handoff when it detects a pilot Ec/Io that exceeds T_add and similarly drops a handoff link if the pilot Ec/Io drops below T_drop ( these being the threshold level settings for adding or dropping a soft handoff). The process of adding or dropping a handoff link does not take place instantaneously, but rather following exchange of some messaging with the base station and after waiting for a pre-defined time threshold to verify the validity of the decision.

The set of thresholds associated with the handoff process represents some of the most challenging parameters to be set for a cdmaOne network. For instance:

o T_add must be set low enough (~ -16 dB) to ensure that handoff is initiated and completed before the signal strength of the original link drops below accept able signal quality. On the other hand, T_add must be set large enough (~ -12 dB) to ensure that handoff does not take place too early, resulting in inefficient utilization of channel elements at the cell site. Therefore, a compromise between the percentage of MSs in handoff and the utilization of channel elements needs to be achieved. It is proposed that on average about 30 percent of the MSs be in soft handoff utilizing two channel elements at different cell sites.
o T_add and T_drop must be set such that an MS does not enter in and out of handoff very frequently. This results in excessive messaging and data processing at both the mobile and base station.
o T_Tdrop must be set to avoid unnecessary dropping out of handoff. This may occur in cases when the MS travels through fast variations in signal strength due to shadowing.
o link balancing: as the cell loading varies, the reverse link will vary in size accordingly. It is required that the
forward pilot link follows to maintain link balance. This can be achieved by varying the pilot power or varying the handoff thresholds so that the handoff window will move accordingly. This also works to avoid creation of what is known as an 'orphan'-a cell with no soft handover into adjacent cells.

In order to reduce unnecessary handoffs, IS-95B introduces a new dynamic threshold on top of T_add and T_drop; it is referred to as T_Dyn-add and T_Dyn-drop and varies based on the current Active Set. Here an MS would wait for the dynamic threshold to be satisfied before sending the PSSM message.

Frequency planning and hard handoff
In a cdmaOne network, the total allocated spectrum is divided into multi-cdmaOne carriers (1.25 MHz for IS-95 systems). Initially, the system is deployed with one carrier and users are assigned to that carrier. As the networks mature and the demand for capacity grows, operators are faced with deploying a second, a third, or even more carriers. It does not make economical sense to deploy multi-carriers through out the whole network; instead, additional carriers can be deployed in areas where the capacity exceeds that capable of being supported on the existing carriers deployed at an acceptable level of quality.

The challenges associated with deployment of multi-carriers are related to the handoff of MSs between carriers as they cross from a multi-carrier region to another or vice versa. In cdmaOne systems, this is referred to as hard handoff between cdmaOne carriers. It is required that an MS entering/exiting a multi-carrier region determines the available carriers, initiates and completes the handoff process in time in order not to drop the call. The system data messaging and processing associated with this process can be extensive.

Many techniques are employed by different infrastructure manufacturers to accomplish smooth hard handoff. One technique employs what is referred to as beacons. These are sectors at the boundary region transmitting only the pilot signal at the other unavailable carrier frequencies. The MS, with the help of the infrastructure, would then switch carriers if needed.

Other techniques propose deploying sectors pointing in opposite directions at the boundaries of the region. By smart-guessing the direction in which the MS is traveling, the infrastructure decides whether or not to instruct the MS to switch carriers. For example, if the MS is determined to be traveling outside of a multi carrier region, it is instructed to switch carriers.

Pilot pollution
In cdmaOne systems, every sector is constantly transmitting its pilot signal as a reference for its MSs to maintain synchronization with the sector. In addition, MSs measure and compare pilot signal strengths to determine potential handoff candidates. A problem arises when an MS is in an area where it can see many strong pilots. Currently, cdmaOne MSs can serve up to three pilots; this causes the extra pilots to act as strong interference or pilot polluters which can lead to reduced system performance in the form of dropped calls and inferior call quality in addition to reduced system capacity.

Some of the techniques one may employ to reduce the effects of pilot pollution are:
o maintain link balancing by reducing pilot power as the reverse link shrinks due to an increase in loading;
o antenna downtilting and rotation; and
o lowering base station heights.

PN offset planning
In cdmaOne systems, all sectors employ the same PN sequence on the forward link-each, however, with a different time shift from the other. This time shift is referred to as the PN offset associated with that sector. All cdmaOne transmissions start at the same instant of time referenced to GPS time. Signals from the different sectors are received at an MS out of sync and the MS needs to only synchronize to its intended sector's PN offset to demodulate its intended signal.

However, due to multipath propagation, variations in propagation delays and propagation loss, three potential problems need to be taken into account in network planning and performance optimization:

o adjacent offset interference: this occurs when two pilots with adjacent PN offsets, due to propagation delay difference, arrive at the MS within a short time shift from one another (within the MS search window). This causes the MS to think they both are a result of multi-path of the same signal and hence combines the two signals leading to a corrupted signal;
o co-offset interference: this occurs when two pilots that are assigned the same PN offset but located at a large geographical distance from one another are received at the MS within the search window. This occurs if the undesired signal is strong enough such as having line of site to the MS. Similar to adjacent offset interference, the receiver would attempt to combine the two signals leading to a corrupted signal; and
o neighbor ambiguity: This occurs when an MS tries to measure the pilot strength of a sector in the neighbor list at a time that another pilot assigned the same PN offset is received strongly at the MS. The MS would then report the pilot strength of the wrong sector and then be instructed to enter handoff with it while it is not the intended sector.

Therefore, in developing a PN offset plan, the designer needs to factor in the effects of propagation delays and signal propagation loss to minimize the potential problems described above.

These, then, are some of the network planning challenges offered by cdmaOne networks today. But the standard is still developing. Currently, an IS-95-based standard is being proposed to IMT-2000 for third generation personal communications systems. 3G systems will support many advanced applications and services such as:

o low-speed voice to high-speed data services ranging from 8kbit/s, 144 kbit/s, 384kbit/s, up to 2mbit/s ;
o packet-switched and circuit-switched services;
o short message services, emails, internet applications, and more.
With these applications come new challenges to CDMA-based techniques:
o faster and more accurate power control algorithms on the reverse link and the forward link;
o improved mobile station receivers combining more resolvable multipath;
o coherent reverse link demodulation techniques;
o multi-carrier planning of wideband width 3G systems overlaid on 2G systems; and
o different design requirements for voice and multi-rate data services.

Clearly, as we move from primarily voice-based services to the true multimedia cdmaOne environment, the challenges of effective planning and optimization will be increased. Products such as MSI's Planet enable operators to model and understand the complex trade-offs necessary to achieve such optimized performance, and further enhancement of MSI products will help to ensure that the needs of the CDMA technology networks of the future will also be fully addressed.