Chapter- 5 Performance Evaluation of Conventional Handoff
Chapter Overview This chapter immensely compares the different mobile phone technologies (GSM, UMTS and CDMA). It also presents the related results for different types of handoff algorithms and different class o f traffic. 5.1. Handoff in Cellular Mobile Phone Networks During a call the mobile station (MS) communicates to the assigned base station (BS) till its signal strength is sufficient. If the received signal strength (RSS) drops below a certain level the call is transferred to a new BS where the RSS is sufficient. This process o f changing the BS or the channel is termed as the hand-off [112]. The hand-off decision is generally taken by the MTSO and executed by the BS. In some cases the MS also participates in the procedure. The set of rules defined for the implementation o f this procedure of hand-off is called as hand-off strategy. There are various aspects of hand-off strategy which influence the performance of the system. Basically RSS is the parameter that takes a major contribution in estimating the requirement of the hand-off. There are other parameters like C/I, BER, etc., also contribute in estimation of hand-off requirement [113]. In practice, a hand-off has a chance to occur anywhere between two adjacent cells. An optimal hand-off decision is one which is made at the right location and time and allows for the best compromise between the number of unnecessary hand-offs and lost calls. No true measures exist for the quality of hand-off algorithm, but a number of situations can be identified where a hand-off procedure shall behave well. Therefore some general demands can be deduced: 1. The number o f hand-offs per covered distance shall be as small as possible. 2. The hand-off shall be carried out as exactly as possible at the planned cell boundary. 3. The target cell shall be chosen correctly. 4. Unnecessary hand-offs shall be prevented by sound decisions.
The procedure and the process of hand-off should be such that the performance measures like blocking probability and forced termination probability should be lowest possible. That is, in a CMCS the hand-off process must be very fast and precise. 5.2 Handoff Algorithm for GSM System The RSS by an MS could be considered as consisting of three components, namely, a free space path-loss component, a slow fading component due to shadowing and a fast fading component due to vehicle velocity. However, when determining handoff necessities, the RSS is averaged and over the normal averaging period, the fast fading component of the signal is averaged out. The shadowing component of the RSS is a function o f cell propagation environment and is a random variable that conforms to lognormal distribution (for 800-900 MHz). Therefore the propagation characteristics of the same environment could be represented by the statistics o f the log-normal distribution. The algorithm for GSM is as follows [114]. 1. The MS performs signal quality measurements on two types o f channels. (a) (b) Measures the RSS and BER of the current forward traffic channel during a call. Measures the RSS of any RF channel which is identified from the measurement order message from the BS/MTSO. 2. The hand-off consists o f three messages: i. Start measurement order. (a) Measurement order message. (b) Measurement order acknowledge message. ii. Stop measurement order. (a) Stop measurement order from BS to MS. (b) MS acknowledgment. iii. Channel quality message. The mobile transmits the signal quality information over the control channel. When a hand-off order is received, if the MS is in low discontinuous transmission (DTX) state then it must enter the high DTX state and wait for 200 ms before taking the hand-off
action. The hand-off action is initiated by the network by turning on a signaling tone for 50 ms then it turns off signaling tone and the transmitter which was operating on old frequency. For the new frequency it adjusts power and tunes to this new channel. After tuning to the new channel the transmitter and receiver is to set to digital mode. Once the transmitter is synchronized, the BS and MS enter the conversation task. 5.2.1. Results for analytical and Simulation Model For GSM network we have evaluated the performance for handoff algorithms based RSS, C/I, and BER. To validate the analytical and simulation model we have tested the model for various handoff algorithms such as - 1. Threshold based. 2. Hysteresis based. 3. Power difference based. 4. Velocity dependent. 5. Multi-criteria base
Performance of Handoff Algorithm in GSM B lo c k in g P ro b a b ility ( P B ) Figure 5.1(a): Validation of Analytical and Simulation Model for PB. probability. Figure 5.1 (a) is the validation for blocking probability and 5.1(b) for dropping
Performance of Handoff Algorithm in G SM Dropping Probability ( PD ) Figure 5.1 (b): Validation of Analytical and Simulation Model for PD From the figure 5.1 we observe that the analytical model and Simulation Model agree within 0.5% difference.
Performance of Handoff Algorithm in G SM Figure 5.2: Performance of Handoff Algorithms for Voice Traffic (Pb) We have performed the comparison of above mentioned handoff algorithms for variable traffic, multimedia traffic and self-similar traffic. Figure 5.2 shows the results for variable voice traffic. From this figure we observe that for threshold based algorithm the new call blocking probability is least and for multi-criteria with HO reservation is the highest.
Performance of Handoff Algorithm in G SM Figure 5.3: Performance of Handoff Algorithms for Voice Traffic (PD) Figure 5.3 shows the performance comparison for dropping probability for the handoff algorithm s in discussion. It is observed that the dropping probability for m ulticriteria is least and is highest for the threshold based algorithm. This due the fact that, for multi-criteria HO algorithm the handoff decision is based on number o f parameters and there is reservation o f the channels for the handoff requests.
Performance of Power Differance Based Handoff for Multimedia Traffic Figure 5.4: Performance of Power Difference Based Handoff Algorithm for Different Class of Traffic Today s cellular systems carry multimedia traffic; therefore we have tested handoff algorithms for such type o f traffic. Figure 5.4 shows the performance of handoff algorithms for voice, data, video, and multimedia (voice plus video) traffic. The blocking probability for multimedia is highest for multi-criteria, moderate for video and least for data only.
It is very important in case of user s interest that the dropping probability should be slightest. We observe that the dropping probability criterion is satisfied by multicriteria HO. 5.3 Handoff Algorithm for CDMA System In case of GSM a hard handoff is performed, that is, the MS releases the old channel before connecting to the new BS via the new channel; hence, there is a short interruption of the connection. Whereas in CDMA systems a soft handoff, in which an MS at the cell border may have several connections to the corresponding base stations at the same time so that there is a smooth transition between the cells without any interruption. To manage the soft handoff between cells belonging to different radio Network Interfaces (RNC). Hence in CDMA additional interconnections between the RNCs are required (in contrast to GSM). In CDMA the handoff decision is based upon the received signal level [115]. A handoff where at every moment the MS is served by the BS from which the maximum signal level is received is called an ideal power budget handoff. Owing to fading effects, such an ideal power budget criterion would cause very frequent forward and backward handoffs between different cells. For an architecture managing soft handoff, there is no problem for switching the connection between the different base stations immediately (on a millisecond timescale); the signals to and from different base stations may even be combined. Because of the short interruption phases and signaling effort, frequent hard handoffs should be avoided. This is usually achieved by introducing an averaging of the signal level and a hysteresis margin, that is, a hard handoff is only performed when the averaged signal level of a neighboring cell exceeds one of the current serving cells by this hysteresis margin of a few decibels [116], [117].
Figure 5.5: Performance of Handoff Algorithms for CDMA Networks (PB) 5.3.1. Results for CDMA Network For comparing the performance o f handoff algorithms in CDMA we have selected to strategies. First one is based on RSS threshold with hysteresis along with soft handoff. Secondly, we add averaging of the RSS prior to hard handoff.
Figure 5.6: Performance of Handoff Algorithms for CDMA Networks. (PD) In case o f CDMA we have evaluated the performance o f the handoff algorithm which takes into consideration assignment o f channel (PN sequence) prior to hard handoff. It is observed that the averaging algorithm gives optimum performance. 5.4. H andoff Algorithm for UMTS System After the successful global introduction o f the second generation (2G) digital mobile communications systems, the third generation (3G) Universal Mobile Communication System (UMTS) has finally taken off, at least in some regions. As we have discussed earlier, CDMA networks can use the same frequency in every cell and users are distinguished by means o f codes. This means that it is a relatively simple task
for a BS to decode the signals from more than one MS simultaneously by dispreading the single received radio signal using a number of different scrambling and channelization codes. This technique is exploited in CDMA to support a feature known as soft handoff, whereby a BS can communicate with more than one MS simultaneously as it moves between cells in the network. The soft handoff has a number of advantages compared with the hard handoff used in FDMA and TDM A systems such as TACS and GSM, in which simultaneous communication with more than one base station is not allowed. In the UMTS soft handoff is controlled by means of an active set, which contains all of the MSs with which the BS is currently communicating. If the active set contains more than one MS, then the BS is deemed to be in soft handoff. On average we might expect around 30% of the BSs to be in soft handoff in a typical network [118]. Although the UMTS system can support soft handoffs, it must also have the ability to support hard handoffs. This type of handoff is required when the MS moves between different CDMA radio carriers or between systems (e.g. moving from UTRA FDD to GSM). This presents a problem to the MS since it must make measurements on a different frequency to assess the suitability of the new radio carrier to support the ongoing call, but there are no natural breaks in the downlink transmissions that would allow the MS to do this. There are two possible solutions to this problem. One solution would be to include two separate radio frequency front ends in every MS. This would allow the MS to make measurements of one radio carrier while still continuing to decode the downlink transmissions from its serving BS on a different radio carrier. While this approach is relatively straightforward from the network point of view, it adds a level of complexity to the terminal that, in many cases, would be unacceptable [119]. Therefore, a second option exists whereby gaps are opened up in the downlink transmissions to give the MS an opportunity to retune to another radio channel and make a measurement. Unfortunately, the amount of data that must flow between the BS and the MS in the downlink direction does not necessarily decrease during the periods when these measurements are required and this means that the BS must transmit at a higher data rate on either side of the measurement gaps to ensure that the same amount of data can be transferred. This mode of operation is referred to as compressed mode because of
the manner in which the data is compressed into the transmission periods on either side of the measurement gap. Figure 5.7: Performance of Handoff Algorithms for UMTS Networks (PB) The purpose o f this strategy is to optimize the cell or set o f cells (i.e. the Active Set) to which the mobile is connected. Although handoff is an inherent functionality with cellular systems, in WCDMA more possibilities are open as long as the mobile terminal can be connected to more than one cell simultaneously, provided that these cells operate at the same frequency. Thus, a distinction between hard and soft handoff can be made.
Regardless o f the handoff type, the handoff mechanism in WCDMA is controlled by the network with the assistance of measurements reported from the terminal side. Handoff involves three different steps: measurements, decision and execution. Measurements carried out by the mobile terminal can be transferred to the network either periodically or they can be event-triggered. The former option may consume radio resources unnecessarily if no changes in the radio interface conditions are observed between consecutive reporting periods. M easurements, which are very precisely Performance of Handoff in UMTS 1 0 ' p i i. r i - j.. e - I W--------------------------------------------------------------------------------- 1-4- 4 4 I --------- I----------------------------------------------1 -------I-------------------------- 1 4-1 t Mobile Assisted SHO L l i j i I P n n v o n t in n a l QUIH Traffic Session/second Figure 5.8: Performance of Handoff Algorithms for UMTS Networks (PD) specified, may be o f different categories: intra-frequency (on the same UTRAN carrier), inter-frequency (on a different UTRAN carrier) or inter-rat. In case o f UMTS the handoff is similar to CDMA system [120]. In this case we have evaluated the performance o f handoff algorithm which takes help o f mobile terminal
for handoff decision. It is observed that the mobile assisted handoff algorithm performs better than conventional handoff algorithm 5.5 Handoff Algorithm for Adhoc Networks Adhoc networks are the ultimate frontier in wireless communication. This technology allows network nodes to communicate directly to each other using wireless transceivers (possibly along multi-hop paths) without the need for a fixed infrastructure. This is a very distinguishing feature of adhoc networks with respect to more traditional wireless networks, such as cellular networks and wireless LAN, in which nodes (for instance, mobile telephone users) communicate with each other through base stations (wired radio antenna) Adhoc networks are expected to revolutionize wireless communications in the next few years: by complementing more traditional network paradigms (Internet, cellular networks, and satellite communications), they can be considered as the technological counterpart of the concept of ubiquitous computing. By exploiting adhoc wireless technology, various portable devices (cellular phones, PDAs, laptops, pagers, and so on) and fixed equipment (base stations, wireless Internet access points, etc.) can be connected together, forming a sort o f global, or ubiquitous, network. Adhoc networks are formed in situations where mobile computing devices require networking applications while a fixed network infrastructure is not available or not preferred to be used. In these cases mobile devices could set up a possibly short-lived network for the communication needs of the moment, in other words, an ad-hoc network. Ad-hoc networks are decentralized; self-organizing networks and is capable of forming a communication network without relying on any fixed infrastructure. Although single-hop ad-hoc networks are often used in practice, when we refer to ad-hoc networks here we always-mean multi-hop ad-hoc networks. The multi-hop support in ad-hoc networks, which makes communication between nodes out of direct radio range of each other possible, is probably the most distinct difference between mobile ad-hoc networks and other wireless communication systems.
Due to the pedestrian semi-static environment the resulting frame duration estimate is in the order of magnitude of milliseconds, and it can be considered low enough to ensure a slow changing environment for the duration of at least a whole transmission frame. We want also to underline that, under the hypothesis of pedestrian environment; the extension of the classical handoff concept to the adhoc networking seems to be meaningful. The handoff, or better, the change of the final sink attributed to a node, is evaluated through the work of a signaling layer. In this way, every node decides, based on a given metric, which is the most appropriate node for its forwarding transmissions. For all these reasons, in an adhoc-networking environment, the handoff concept is replaced by the forwarding neighbour selection concept. This reflects even more the prerequisite of intelligence shift (from the base stations to the terminals) that is necessary to make the adhoc architecture performing properly Above stated hypothesis implies that in case of adhoc networks the blocking probability will always be zero. That is every packet will find some neighbour for forwarding the packet. Hence we evaluated the dropping probability. Figure 5.9: Performance of Handoff Algorithm in adhoc Network Figure 5.9 shows the dropping probability (in case adhoc network we have estimated packet dropping probability) for three type of handoff procedures. These handoff procedures are somewhat different than handoff procedures in cellular networks.
Figure 5.9: Performance of Handoff Algorithm in adhoc Network 5.5 Concluding R em arks In this chapter, we have performed the performance evaluation o f Four type of networks, which represent almost all classical wireless networks. Firstly we have validated the analytical and simulation model which was suggested in chapter 3 and chapter 4. Further investigations were carried out for different type o f handoff algorithms and different class o f traffic.