Handoff Algorithms in Dynamic Spreading WCDMA System Supporting Multimedia Traffic

Transcription

1 Handoff Algorithms in Dynamic Spreading WCDMA System Supporting Multimedia Traffic Ju Wang, Jonathan C.L. Kavalan PRESENTED BY: KUNWARDEEP SINGH GAYATRI BEHERA

2 Introduction Multimedia data traffic is more demanding than voice service, both in data rate and maximum tolerated bit-error rate (BER), which need to be guaranteed at all times In this paper, integration of soft handoff and dynamic spreading factor in wideband CDMA system in supporting multimedia traffic is discussed Contribution of this work is: Proposed a batch mechanism such that the average delay is reduced from at a heavy handoff rate A new resource allocation algorithm is introduced, where the spreading factor and transmission power for the handoff mobiles are jointly considered to maximize the throughput

3 Terminologies discussed CDMA Handoff Spreading Factor (SF) Bit Error Rate (BER)

4 Background and Related Study CDMA supports soft handoff scheme having no connection break during handover. Traditional Handoff algorithms focused on voice only network with guaranteed connectivity. Radio resource allocation has been discussed in previous works. But they assumed fixed SF and didn't address resource assignment for reverse link. Works that involved optimization by reallocation of radio resources and dynamically changing SF were not reported before.

5 Proposed Scheme The basic functionality of the proposed handoff scheme has three parts: 1. Collect information of the surrounding cells and the mobile profile 2. Update the spreading factor (SF) in the active set (BSs in range) to maintain BER level for mobiles in these cells 3. Decide the right SF and power level for handover mobiles.

6 Proposed Scheme(contd.) SF and power level decided such that following conditions are satisfied: 1. Link quality of mobile stations in the neighboring cell isn't reduced. 2. Minimum BER requirements is not violated. 3. Maximum data rate of handovering mobile is achieved.

7 Proposed Scheme(contd.) Problem: Handoff algorithm consumes high system resource, results in long handoff delay Solution: Handovering mobiles need to be processed in batch. Update operations caused by consecutive handoff should be combined together. Delay reduced by almost 29%.

8 Proposed Scheme(contd.) A suboptimal method (HYB) is proposed with goal of joint optimization for SF and power level for all mobiles in handover area The algorithm models BER constraints of all mobiles as the function of various factors such as: (SF, power) pair for handoff mobile System load of active set Distance from active cells

9 Framework of Soft Handoff Algorithm with Dynamic Spreading

10 Performance goals of the handoff algorithm with dynamic spreading factor All handoff requests should be processed within given time period, so that mobile can smoothly migrate to other cell. Algo should be able to handle stress handoff requests, when system under heavy load. Handoff algorithm should seek highest throughput while maintaining the BER performance of handoff mobile during handoff period.

11 Stationary System Behavior The handoff algorithm needs to decide: whether relevant active set should perform an update process whether handovering mobile should change its SF Probabilities for 4 situations are derived based on the updating decisions for mobile and target cell: [mobile unchanged, cell unchanged] [mobile unchanged, cell update] [mobile update, cell unchanged] [mobile update, cell update]

12 Stationary System Behavior (contd.) Results show that 63% of the time there will be an update for handoff mobile, unchanged in the target cell. The probability of no update in both sides is only Thus, proposed handoff scheme could lead to considerable amount of updates if handoff occurred.

13 Handoff Time Analysis Handoff processing time is defined as the total time between a mobile entering into handoff state and the user become ready to communicate with the active set T handoff = T i + T u + T e T i = time to exchange information among the active set T u = time to evaluate the impact to the target cell caused by the handoff mobile. Decide if update in the target cell should be performed. T e = time to choose the new spreading factor, signal power, update time of the handoff mobile.

14 Handoff Time Analysis(contd.) Handoff time under stressed attack Improved handoff time using batching algorithm

15 Determining SF for Handoff Mobile In the handoff area, system should decide 2 parameters: spreading factor transmitting signal strength If decision of SF and power level is separated, either the link quality gets degraded or radio efficiency is decreased In Maximize Data Rate scheme (MaD), shortest possible SF is chosen When MaD is coupled with power control schemes, either BER requirements are failed or interference is high An optimal approach is required which is discussed later

16 OR-ON-DOWN Power Control A close-loop power control used in IS-95. BSs in control set issues power control command to mobile based on received signal strength UP command means that the mobile should increase its transmission, DOWN command means to do the opposite Works well for voice traffic but not with dynamic SF BER exceeds the allowed value

17 Main Path Power Control Main path: Radio channel between mobile and cell allowing the shortest SF Signal of at least one radio link should be strong enough to provide required BER. Cell in the main path carries out the power control during hand-off period. However, the strong signal of the handoff mobile also contribute to the interference level for other users and BER limit violations happen

18 Performance Optimization in Handoff Period An open loop SF and power control algorithm is proposed based on accurate estimation of link BER for both the handovering mobile and cells in active set. Approximate of channel BER can be expressed as a function of E/I(SIR, or energy-per-bit-to-interference ratio). SIR for the jth handoff mobile can be expressed as:

19 Performance Optimization in Handoff Period(contd.) Overall throughput is sum of B c /SF j for all handoff mobiles B c is chip rate of system (5 Mc/s for WCDMA) Optimal SF/Power assignments can be obtained by solving the following nonlinear problem:

20 Simplification of Original Problem: Optimization problem is reduced to linear programming problem which can be solved by the popular simplex method. Unfortunately, optimal solution only provides an upper bound to the original problem which cannot be reached at times. Optimal power vector could contain zeros for some P j. Also, according to constraints, SF j P j must be greater than a positive value, which is impossible if P j = 0

21 Proposed Suboptimal SF/POWER Decision Algorithm: The algorithm is named HYB which utilizes the linear programming methods and additional constraints reevaluation from the original problem. First, Optimal power vector is found using the standard simplex. Check if there are any P j = 0 or SNR below threshold. If no, obtain corresponding SF j and further process it to result in a larger SF j and repeat to find P j, which will be less than original P j. Terminate when we have all decision variables. Otherwise, transform to further simplify and generate new problem with reduced decision variables. Algorithm will eventually terminate and the maximum loop time will be less than the number of handoff mobiles

22 Numerical Results and Performance

23 Numerical Results and Performance (Contd)

24 Observed Improvements: For WWW traffic experiment, most of the BER lied between to The close-to-desire BER performance showed that system resources were shared equally For video handoff traffic experiment, BER lied between and So, even the worst case of HYB can benefit from 26% performance gain Reduction in average delay from 1.12 s to 800 ms during heavy handoff.

25 Conclusion Update process in the dynamic spreading was revised to prefer batch processing such that overall delay is reduced, as well as the number of updates. SF and transmitting power level for handoff mobiles are optimized to maximize throughput Problem formulated to a nonlinear programming form and a heuristic algorithm to find the suboptimal solution was proposed Overall 25% increase in throughput for WWW traffic and 26% for video traffic illustrated by numerical results