DOC PREVIEW
Berkeley ELENG 228A - Improving Load Balancing mechanisms in Wireless Packet Networks

This preview shows page 1-2 out of 5 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 5 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 5 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 5 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

Improving Load Balancing mechanisms in Wireless Packet Networks1 Giuseppe Bianchi, Ilenia Tinnirello Dept. Electrical Engineering, University of Palermo Viale Delle Scienze, Parco D'Orleans, 90128 Palermo, Italy This work was supported in part by MURST. Abstract - This paper provides a comparative performance evaluation of various load balancing schemes in cellular packet networks. With respect to circuit switched networks, wireless packet technology adds the further issue of Quality of Service of accepted connections. In fact, with packet technology, transmission error performance does not uniquely depend on the perceived channel quality, but it can be improved by adopting a scheduling mechanism enforcing fast retransmission of corrupted packets. The result is that throughput can be traded off with QoS experienced by an admitted flow. This paper proposes new packet-level load balancing mechanisms. In addition to the number of calls admitted in a cell, our schemes use supplementary packet level information, expressed in terms of effective resource consumption of each individual call when retransmission mechanisms are employed. Simulation results prove the superiority of our proposed schemes with respect to traditional load balancing schemes. I. INTRODUCTION Traditional frequency and time division cellular systems are based on fixed channel assignment. In such systems, a fixed channel resource, e.g. a TDMA slot per frame, is allocated to an admitted call for its whole duration time. It results that the QoS offered to the users is summarized in two distinct and independent performance figures: transmission quality, which depends on the transmission error probability encountered as well as on the specific error correcting coding employed, and outage probability (i.e. the probability that a call is refused or subject to a forced interruption) . Load balancing algorithms [1,2,3,4] have been considered to improve the outage probability in wireless cellular networks. These algorithms operate when the coverage areas of different Base Stations (BS) overlap. Whenever a Mobile Station (MS) can attach to more than one BS, the idea is to direct a new call to the BS with the greatest number of available channels. It has been proven that this operation minimizes the probability that future incoming calls (either newly originated or handovering ones) will be blocked because of lack of resources. Wireless networks based on packet technology and dynamic resource assignment [5,6,7,8] are increasingly popular. In these systems, channel access is dynamically controlled by a packet (slot) scheduler running at the BS, which grants channel access to the MS with most urgent need to transmit. Dynamic resource assignment is particularly effective as it provides the capability to retransmit, in a very short time frame, corrupted packets, thus improving the packet error rate in the presence of critical channel conditions [5,6]. Emerging commercial WLAN systems, such as Bluetooth, and the Point Coordination Function of the IEEE 802.11, have included dynamic packet scheduling in their channel access control operation. Fast ARQ is also envisioned in third generation cellular systems (UMTS). The thesis carried out in this paper is that, in wireless packet networks, load balancing algorithms should be designed to take advantage of additional "packet level" information, rather than be limited to "call level" information (number of accepted calls per cell). In fact, in dynamic resource assignment schemes, the transmission error probability, i.e. the packet level QoS, is no more independent on the cell occupancy status. On the contrary, packet retransmission capabilities imply that the greater the amount of spare channel resources, the larger is the number of retransmission that can be attempted, the better the transmission quality provided. Scope of this paper is to propose novel load balancing schemes driven by packet level information, and prove their superiority with respect to traditional load balancing approaches. More into details, we propose and evaluate the performance of two novel metrics, based on packet level information, designed to drive the BS selection algorithm for newly originated or handovering call. The paper is organized as follows. In Section II, dynamic resource assignment systems are briefly described. Section III proposes and motivates the novel BS selection metrics. In section IV, a performance evaluation is carried out by means of simulation. Finally, conclusive remarks are drawn in section V. II. DYNAMIC RESOURCE ASSIGNMENT In a micro cellular environment, the small propagation delay allows an efficient command/response communication mode. This feature has been effectively used in several recently proposed multiple access control (MAC) schemes, which base their operation on dynamic resource assignment [5,6,7]. In these schemes, the channel is divided into fixed size slots. However, they differ from fixed TDMA access mechanisms as each slot is not uniquely reserved to an attached connection, but is dynamically assigned by a central scheduler placed on the BS. Dynamic assignment is implemented by a command-response protocol, either on a slot-by-slot [5,6] or a frame-by-frame [7] basis. Dynamic resource assignment schemes appear effective when multi rate and/or variable rate traffic sources, with different delay requirements and priorities, need to share the same medium. However, it has been proven [5,6] that a flexible slot assignment turns into an effective performance enhancement. 8910-7803-7400-2/02/$17.00 © 2002 IEEE.12345145412345673Slots used for first tx:1st cycleSlots used for retransm:2nd cycle 3rd cycleA)B)Slots used for first tx:1st cycleretx:2nd cycle Fig. 1. Graphical illustration of the dynamic resource assignment scheme adopted. Each connection is identified by a different number. Unsuccessful transmissions are marked. The idea is that idle slots can be used to retransmit corrupted packets: the greater the maximum packet delay and the spare capacity, the greater the probability of


View Full Document

Berkeley ELENG 228A - Improving Load Balancing mechanisms in Wireless Packet Networks

Documents in this Course
FAST TCP

FAST TCP

57 pages

Load more
Download Improving Load Balancing mechanisms in Wireless Packet Networks
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Improving Load Balancing mechanisms in Wireless Packet Networks and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Improving Load Balancing mechanisms in Wireless Packet Networks 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?