The Performance Impacts

Although the compressed mode could help on the inter-system handover, some system performance besides of the signal quality (such as Bit Error Rate or Frame Error Rate) will also be impacted:

¾ Coverage

Because more power is needed for the same QoS constraint during the compressed mode, the maximum path loss budget is reduced. As a result, the distance that the mobile can transmit is reduced, thus the uplink coverage is reduced. In [5], the path loss will be reduced by 2.3dB if compressed frame is operated every second frame.

¾ Capacity

Since the power control cannot work during compressed frame, the transmitted data needs higher Eb/N0 requirement to maintain the quality. The capacity will be decreased by the higher Eb/N0 requirement. The capacity can be reduced by about 2% even only 10% of users are operating the compressed mode every third frame [5].

¾ Code Shortage Problem

When the “reducing the spreading factor by two” method is applied, the blocked code is twice than before and the available code space is reduced. There might not have enough codes to support all active users especially for high data rate users (low spreading factor). There are three proposed solutions for the code shortage problem: 1.

Using Non-orthogonal scrambling codes. 2. Avoid the transmission gaps from different mobiles overlap in one frame, the spreading factor would be used in fair distribution [23].

3. Base station assigns a dynamic common channel to a mobile station performing inter-frequency/system handover [24].

Here, the challenges are how to find a robust mechanism to get the best efficiency and performance by allocating the transmission gaps in various fading and the relative uplink/downlink power control before and after the transmission gaps for different fading condition. It is critical to have an adaptive compressed mode operation to minimize the impacts on system performance.

Chapter 3 Capacity-Based Compressed Mode

The performance impacts of the compressed mode are studied. To reduce the impacts and maintain the compressed mode efficiency, a capacity-based compressed mode is proposed in this chapter. The concepts and the implement of the proposed control algorithm are discussed here.

3.1 The Prior Works

So far, many articles discuss the compressed mode performance. However, they mostly modify compressed mode algorithms to enhance the border-cell handover success rate without concerning the potential impacts caused by the compressed mode. Besides of three compressed mode methods suggested by 3GPP standard [4], Gustafsson et al. [18] provide the formula of the transmission bit rate and relative parameters to generate the transmission gap.

Moreover, the issues of triggering criteria for the compressed mode are researched. Ying et al.

[25] compare the periodic and event-triggered compressed mode. With extra overhead cost, the period-triggering algorithm has higher handover success rate. Zhang [26] considers the base line quality and suggests that the pilot Ec/Io is suitable for high traffic load and the pilot received signal code power (RSCP) is suitable for low traffic load. The combined pilot Ec/Io and pilot RSCP triggering method is then proposed to guarantee good system performance under all traffic loads. The relationship of the compressed mode gap pattern and GSM carriers measurement are considered in [22][27]. The required handover time for different transmission gap patterns is studied.

During each compressed frame, more power is suggested to guarantee the quality of increased transmission rate. The interference caused by the increasing power will impact the performance in either capacity or coverage. Holma and Toskala [5] quantify the degradation of the capacity and coverage. Some discussions in 3GPP TSGR4 meetings also address the impact scenarios [28-30]. Although the above articles point out the performance impacts, no proper solution is proposed. To resolve that, a capacity-based compressed mode is proposed

to balance the tradeoff between the handover success rate and the capacity.

If the mobiles in the border of UMTS cells want to hand down to GSM cells, it uses the compressed mode to measure GSM carriers. However the GSM measurement is often not enough in current implementation. The critical problem is that measure at wrong time when GSM doesn’t transmit signals. With the wrong time measurement, not only measure no signal but also waste power without any profit. It prefers that all mobiles measure at the right time simultaneously and then use the proposed capacity-based compressed mode to limit the interference level. It benefits both effective measurements and performance maintenance.

3.2 The Concept of the Capacity-Based Compressed Mode

The increasing power is to compensate the influence of the transmission gaps, but it also impacts the performance on capacity or coverage. In this thesis, the fixed cell size is calculated by the maximum path loss, so only the capacity impacts are considered. To calculate the uplink capacity, the required bit energy to interference ratio (Eb/I0) can be computed in equation (3-1). For simplicity, the same traffic services and the same received power at the base station are assumed for all mobiles.

] where Eb is the received bit energy, Io is the total interference, S is the received power at the base station, PG is the processing gain, F is the noise figure, Nth is the thermal noise power density, W is the transmission bandwidth, α is the voice activity, β is the adjacent cell interference factor, N is the number of users, NCM is the number of users in the compressed mode, and ΔS is the average increasing power for the compressed mode. The capacity, Nc, can then be calculated in equation (3-2), where

et

⎛ is the target bit energy to interference ratio. The capacity will be deducted by

⎟⎟⎠

The more users in the compressed mode the more capacity will be degraded. So a capacity-based compressed mode is proposed to reduce the impact.

According to equation 3-2, Figure 3-1 shows the capacity with varied interference. The capacity with the compressed mode is obviously lower than the capacity without the compressed mode. Initially the scenario of many users with the compressed mode is located at point a. The interference exceeds the maximum tolerated value and the capacity is only CCM. In order to enhance the capacity, the number of users with the compressed mode is reduced and the relation curve tends to the non-compressed mode curve gradually (the operating point towards b). The interference is then reduced and the capacity can increase toward CNCM (as point c). Either the number of users in the cell is reduced (as point d), and the interference is low enough to restart the suspended compressed mode. With the suspended compressed mode users operating the compressed mode again, the operating point tends toward point e. Last, when the number of users increases, the operating point backs to point a. This plot announces that arranging the compressed mode operation can effectively improve the capacity.

Number of users

Figure3- 1. The capacity with different interference

When concerning about the downlink capacity, the downlink capacity is decided from the maximum power budget as equation (3-3):

max where Ptot is the total base station transmitted power, POH is the overhead power, N is the number of users in the cell, α is the voice activity, Pi is the base station transmitted power for single user, Mi is the multiplier of compressed mode power increase, and Pmax is the maximum power budget. The Pi is decided by the target bit energy to interference ratio

et when the compressed mode is not operated. When operating the compressed mode Mi is equal to the multiple of power increase. The total transmitted power is required to be less than the maximum power budget. It can be observed that the more power increases in the compressed mode the less capacity could be supported. Whether the uplink or the downlink, the capacity is decreased when the number of the compressed mode users is increasing. Consequently, a capacity-based compressed mode is required for both uplink and downlink transmission.

3.3 The Compressed Mode Format

The compressed mode format is composed of gap generation method, triggering criteria and gap pattern. In the simulation, these factors are considered as below:

a. Gap generation method

As discuss in chapter 2: Puncturing cannot be used in uplink, and can only generate small gaps. Scheduling cannot apply to real time service. Only reducing the spreading factor by two is chosen into simulation among the three methods and can be suitable for all 3G services.

b. Compressed mode triggered criteria

The pilot Ec/Io is the typical threshold for the UMTS soft handover. However the pilot Ec/Io is not suitable for the compressed mode triggering due to following two reasons. The first reason is the border cell effect which means the Ec/Io decays smoothly when the mobile away from the base station. In border cells, the pilot signal and the interference are under the same fading condition, thus the pilot Ec/Io will hold the value until hitting the noise limit, where the total interference is dominated by the background noise. Figure 3-2 shows the different Ec/Io degrade scenario between the center cell and the border cell. It can obviously see the curve keeps consistence in the border cell and the pilot Ec/Io can not reflect the signal strength. The border cell effect is also illustrated in [31] and the curve also follows the same trend. The second reason is that the pilot Ec/Io is influenced by the different loadings. Figure 3-3 shows the pilot Ec/Io varies with different loadings. In the center cell, the UMTS uses relative threshold to trigger handover. However in the border cell, the mobile might measure only one pilot signal. As a result, the algorithm hardly finds a proper absolute threshold adapt to all the different loading. In this case, the pilot Ec/Io is not suitable for being the compressed mode triggering and handover decision.

To ensure in-time measurement and avoid unnecessary compressed mode triggering, the event triggering by pilot RSCP is chosen. For a baseline performance of measurement, the simulation starts the compressed mode when the pilot RSCP is smaller than -95 dBW and stop the compressed mode when the pilot RSCP is larger than -90 dBW. If the mobile keeps on going out, it will hand-down to GSM cells. The handover triggers when the RSCP is smaller than -118 dBW for 500 ms. The trigger timer is used for avoiding the ping-pong effect. The above parameters are listed in Table 3-1. The threshold settings are according to the distance from the base stations with zero fading.

Table3- 1. Compressed mode triggering threshold

Threshold Value Distance

Compressed mode start threshold for pilot RSCP -104 dBW 0.5*radius Compressed mode stop threshold for pilot RSCP -108 dBW 0.6*radius Handover triggering threshold for pilot RSCP -118 dBW radius

Time to trigger handover 500 ms

Figure3- 2. The pilot Ec/Io decay curve in center and border cells

Figure3- 3. The pilot Ec/Io distribution function with different londing c. Gap pattern to measure GSM carrier

In GSM, only Frequency Correlation Channel (FCCH), Synchronization Channel (SCH) and Broadcast Channel (BCH) are transmitted at all time. To be useful, the measurement of GSM carriers and Base Station Identity codes (BSICs) [14] should only on FCCH and SCH.

However, the gap patterns in UMTS specification [14] are not guaranteed to match with the GSM timing structure. Ideally, the measurement gap should be 9.2ms for every 46ms, but the formula does not match UMTS format. In the simulation, the gap time of 14 slots for every 5 frame (9.3ms for every 50ms) is chosen to approach the GSM control frame structure. The GSM channel scenario is depicted in Figure 3-4(a) and the corresponding gap pattern is depicted in Figure 3-4(b).

F S BBBBP P P P S P P P P P P F PPFS P P PPPPPPFS P P PPPPPP F S P P P P P PPPX (a)

(b)

46ms 46ms 46ms 46ms 50.6ms

Figure3- 4. (a) GSM control channel (b) Compressed Mode gap pattern

3.4 The Algorithm of the Capacity-Based Compressed Mode

When a mobile needs to measure other systems such as GSM, the mobile can measure only at few measurable time slots. As a result, all the users in border cells will execute the compressed mode simultaneously to match the measurable time slots as depicted in Figure 3-5(a). The increasing aggregate power could cause a serious impact on the capacity. To resolve the power problem, two methods are suggested and are depicted in Figure 3-5(b) and 3-5(c). The first one is to separate the position of transmission gaps and spreads out the aggregate increasing power. The second one is to schedule the order of the execution of the compressed mode. However, the first one moves the transmission gap to the adjacent time slot but there is no guarantee that the new measurement interval can match with the actual transmission slots of other systems. As a result, the second method is chosen for the proposed algorithm.

Figure3- 5. The compressed mode scheme (a) Normal compressed mode at simultaneous time (b) Separate the position of transmission gap (c) Schedule to suspend the compressed mode

First, the relationship of pilot RSCP versus the distance from the base station is depicted in Figure 3-6. When the mobile is close to the base station, the curve shows the exponential increase of the RSCP. When the mobile is away from the base station, the curve tends to stay linear. By using the linear relationship, the critical pilot RSCP ratio, RRSCP, to represent the distance that needs performing the compressed mode is calculated in equation (3-4).

ho stop stop

RSCP T T

RSCP R T

−

= − (3-4)

where Tstop is the threshold to stop the compressed mode, and Tho is the threshold to trigger border-cell handover. According to the previous compressed mode format, the compressed mode operates only in between Tstop and Tho (RRSCP is ranged in between 0 to 1). The compressed mode stops when RRSCP equals to 0. The hand-down to GSM occurs when RRSCP

equals to 1. The relationship between RRSCP and RSCP is shown in Figure 3-7. According to RRSCP, it can estimate the proportion of the effective distance for the compressed mode operating.

Figure3- 6. The relationship of Pilot RSCP and distance

Tstop Tstart Tho

RSCP Status

Figure3- 7. The relationship of RRSCP and distance RRSCP

Non-compressed mode Compressed mode

0 1

The proposed capacity-based compressed mode limits the power level in the compressed frame by suspending low priority users from operating the compressed mode. Thus it can ensure the capacity while maintaining the proper hand-down priority. The following steps of this algorithm are described as follows:

(1) Every frame, the suspend factor Fi(n) base on the pilot RSCP ratio RRSCP, the aggregate measured GSM samples Nmeas(n), and the record of last time compressed mode Rsuspend is calculated in equation (3-5).

As shown, the algorithm prefers to suspend the users close to the base station (small RRSCP). The users keep larger sampled GSM carriers, Nmeas, will easily be suspended. The schedule algorithm wants to make each user have the same

k RSCP

meas

R

N ; it can balance the aggregate measured samples with the effective distance for the compressed mode operating. Besides, there is a tunable factor k to modify the schedule algorithm. The factor k modifies the dominant degree of the distance. Finally, the Rsuspend equals to 1 when the last time compressed mode operation have been suspended, otherwise it equals to 2 as equation (3-6). This term halves the suspend priority for users who just has been suspended and is designed to reduce the chance of the continuous suspension of the same user. The continuous suspension may delay the measurement efficiency and it might be delayed to handle the emergency handover.

(2) The second step is to observe the base station transmitted power and set a suspend threshold of transmitted power Pthr, which is smaller than the maximum transmitted power budget. If the estimated base station transmitted power, PBSest, doesn’t exceed the threshold Pthr and then the system can operate the compressed mode without any suspension to guarantee the measurement efficiency. If the PBSest exceeds Pthr, then the system suspends the compressed mode according to the priority order based on Fi(t) until the transmitted power below Pthr or there is no other compressed mode available.

The flow chart of the capacity-based compressed mode is depicted as Figure 3-8. The base station collects the information of the compressed mode and computes the suspend factor for

all users in the compressed mode on every frame and schedule the users to suspend their compressed mode operation. Then, the base station examines the transmitted power and decides whether to suspend the compressed mode. As a result, the proposed algorithm reduces by suspending better RF users to maintain the system capacity.

Figure3- 8. The flow chart of capacity-based compressed mode

In this study, the proposed control algorithm will consider only the downlink capacity.

The reasons are as follows:

1. Only the base station can acquire the information of all users’ compressed mode profiles.

The base station executes the capacity-based compressed mode and sends messages to all mobiles in the cell to control the compressed mode.

2. In the asynchronous data transmissions, the required throughput of the downlink is much higher than the uplink. As a result, the capacity limit is on the downlink in the multimedia service. The downlink transmission is required to propose a control algorithm to take care its performance.

3. In most cases, the compressed mode is used in the downlink. Since the base station can stops the transmission and let the mobiles have the free time to measure the signal strength of other systems. If both the downlink and uplink compressed mode is performed, the addition time-aligned with time offset is needed.

Compute Suspend factor with users in compressed mode

P_BSest > P_thr ? N

Y

Next Frame

Schedule the priority user for the compressed mode suspension

New Connection

Other users in the

compressed mode? N Y

Chapter 4 Simulation Platform

The proposed capacity-based compressed mode has been introduced. It needs a simulation platform to verify the performance. This chapter introduces the simulation platform and the platform is constructed of 19 UMTS cells. The cell radius is calculated by the uplink link budget and the propagation model. The details of the environment establishment are addressed here. Moreover the capabilities of the platform are also presented.

4.1 Simulation Environment and the Mobility Model

This simulation model constructs UMTS wireless communication system scenarios.

There are 19 UMTS cells each with 3 sectors and the UMTS is surrounded by GSM Sea as depicted in Figure 4-1. Each cell has three antennas and the angles of the antennas are 0°, 120°, and 240°. In each UMTS sector, the base station will execute the soft handover and downlink power control to support the mobility. When the moving mobile is away from the serving sector, the signal strength is weakened and the connection quality is degraded. The soft handover connects to other sector to guarantee the seamless connection, and the downlink power control ensures the quality of the received signal to interference ratio (SIR).

UMTS

GSM

Figure4- 1. The Simulation Platform

This platform can simulate with arbitrary numbers of mobile within the supported capacity in each sector. It only observes the mobiles in the border sector. Initially, all of the mobiles are random placed in the border sector with uniform distribution. Each mobile is

This platform can simulate with arbitrary numbers of mobile within the supported capacity in each sector. It only observes the mobiles in the border sector. Initially, all of the mobiles are random placed in the border sector with uniform distribution. Each mobile is