TRAFFIC CONTROL ALGORITHMS FOR A MULTI ACCESS NETWORK SCENARIO COMPRISING GPRS AND UMTS

---

Presented By:
Shaheen, B.Tech ¾ E.C.E
Shadan College of Engg and Tech,
Hyderabad
T. sumalatha, B.Tech ¾ E.C.E
G.Narayanamma Institute of Tech,
Hyderabad,

---

ABSTRACT
Next generation wireless communication systems will bring new architecture concepts, new spectrum allocation policies and smart resource management in a multi-network environment. The mobile and wireless devices of the future will be more powerful, less heavy, and comprise new interfaces to the user and to new networks. The topics are based on the paradigm of providing improved services to the end user in wireless communication networks.
This paper assumes a multi-network scenario comprising GPRS and UMTS Radio Access Networks (RAN) and proposes several schemes for controlling and distributing the network traffic over these two RANs, increasing the spectrum efficiency of the whole system by ensuring that Mobile Terminals are served by the optimum radio access technology available. This is achieved by assigning a cost to each service that the network can provide, taking into account both system load and system resource availability, and enabling the Mobile Terminal to select the most efficient RAN, retaining the quality the user asks for. The results show that a gain, in terms of percentage of satisfied users, of more than 15% can be obtained by the multi-network system compared to the case of no traffic distribution process, thus leading to a more efficient spectrum usage.
INTRODUCTION
Multi radio access systems are likely to dominate the forthcoming mobile communication scenario. In fact many network operators that now are running only one RAN, with the roll out of UMTS will manage both GSM (GPRS) and UMTS networks. Moreover, in the future co-operation among different operators
is envisaged, by means of legal agreements, in order, for example, to make a shared and optimized use of the spectrum resource. In this scenario also broadcast technology is expected to be involved into data delivery service.
Nowadays every radio access system is developed to exploit in the most efficient way the available resources to deliver its set of services, but if a multi-network system run by one operator is envisaged, the necessity of best exploiting the overall available resources arises by all available RANs. Multi-network system architecture has been proposed by DRIVE and it is further studied in the ongoing OverDRIVE project. One of the key aspects in increasing spectrum efficiency in a multi-RAN system is Traffic Control (TC). Assuming that the same service can be delivered by more than one RAN with the same quality, Traffic Control should answer the question: which is the most suitable access network the new user should be assigned to, so that the overall spectrum efficiency would be the greatest? This is the objective of the Traffic Control algorithm which selects the optimum RAN taking into account the actual conditions of the networks in terms of system load, resource availability and resource consumption the Mobile Terminal (MT) would cause. The TC schemes presented in this paper are based on a service cost definition, which reflects the above characteristics. Each RAN calculates the cost associated with each service it can provide and then broadcasts these costs. The MT receives this information and makes the decision. It can be noticed that this approach does not require any interface definition among different RANs, as the MT is the only point of contact between different RANs.
APPLICATIONS
Tomorrowâ„¢s cars will comprise many wireless communication and mobility aware applications. Music, News and road conditions weather reports and other broadcast information is received via digital audio broadcasting with 1.5 Mbits/sec. for personnel communication, a global system for mobile communication (GSM) phone might be available offering voice and data connectivity with 384Kbits/sec. for remote areas satellite communications can be used, while the current position of the car is determined via global positioning system. Additionally, cars driving in the same area build a local ad hoc network for fast information exchange in emergency situations or to help each other keeping a safe distance. Incase of an accident, not only will the air bag be triggered, but also an emergency call to a service provider informing ambulance and police.
Future cars will also inform other cars about accidents via the ad hoc network to help then slow down in time, even before a driver can recognize the accident. Networks with a fixed infrastructure like cellular phones (GSM, UMTS) will be interconnected with trunked radio systems (TETRA) and wireless LANs. Additionally, satellite communication links can be used. The networks between cars and also inside a car will more likely work in and ad hoc fashion. Wireless Pico networks inside a car can comprise PDAs, laptops and mobile phones, e.g., connected with each other using the blue tooth technology.
EMERGENCIES: after an accident has occurred, vital information about injured persons can be sent to the hospital immediately. There, all necessary steps for this particular type of accident can be prepared are further specialists can be consulted for an early diagnosis. Wireless networks are the only means of communication in the case of natural disasters such as hurricanes or earthquakes. In the worst cases only dec3ntralised, wireless ad hoc networks survive.
BUSINESS: todayâ„¢s typical traveling salesman needs instant access to the companyâ„¢s database to ensure that files on his or her laptop reflect the actual state, to enable the company to keep track of all activities of their traveling employees, to keep databases consistent etc. with wireless access, the laptop can be turned into a true mobile office.
REPLACEMENT OF WIRED NETWORKS: wireless networks can also be used to replace wired networks, as for remote sensors, for trade shows, or in historic buildings. Due to economic reasons it is often impossible to wire remote senses in cases such as weather forecast, earthquake detection, or environmental information. Wireless connection via satellite can help this situation.
INFOTAINMENT AND MORE: wireless networks can provide up to date information at any appropriate location like on history of some building, paying via electronic cash, another field lies in entertainment and other games in order to enable e.g., ad hoc gaming networks as soon as people meet to play together.
MOBILE AND WIRELESS DEVICES: there is no precise classification of such devices, by size, shape weight or competing power. Currently, laptops are considered to be the upper end of the mobile device range. The following list gives some examples of mobile and wireless devices graded by increasing performance.
Sensor
Embedded Controller
Pager
Mobile Phones
Personal Digital Assistant (PDA)
Palm top or Pocket Computer
Notebook or Laptop
Traffic Control Concept
Consider the case of an operator running both a GPRS and a UMTS network: for new high data rate services, an increasing number of users will subscribe to UMTS but speech and low data rate services will still be offered by both GPRS and UMTS. It is clear that providing a service by means of GPRS, which has a limited number of channels, or by means of UMTS, which has a limited Base Station power, implies differences in spectrum efficiency, which is influenced by many factors: the load of the network, the increase in the load due to the provision of that service and the available resource. Since efficiency of providing a service may differ from system to system, the overall efficiency can be increased by introducing a TC functionality that selects the most suitable radio access system.
A TC system function has been proposed within the IST DRIVE Project. For each service and QoSs the associated cost is broadcasted to the cell. Note that the term â„¢costâ„¢, used here, does not refer to any financial costs, but is a weighting factor determined from the traffic conditions in the network. The MT receives this information from each RAN it is able to listen to and then, according also to the user preferences concerning quality and billing, the choice of the system which offers the desired service at the lowest cost is made. The key aspect in this approach is that the RAN selection is made at the MT side. This implies that different RANs do not need to exchange information, but they have only to agree a common definition of the QoSs and of the costs to assign to each service. This is also compatible with a scenario in which the access networks are operated by different companies and therefore would not know the conditions on the other RANs. Then different technologies merge into a common point that is the MT.
Traffic Control Algorithms
This work investigates cost-based TC mechanisms in a scenario comprising a GPRS and a UMTS network providing cellular coverage in the same area. The networks calculate a cost for each cell in the RAN and the MTs requesting a service are always given the most up to date cost information since it is re-calculated each K = R time the networks goes through major changes. Other algorithms that update the cost on a periodical basis can be found in [6].The purpose of the cost is to assign the user to the RAN to which it causes the smallest resource consumption and the smallest decrease in capacity, in terms of number of users that can be further accepted. GPRS and UMTS provide different types of resources, there-fore the cost is in general defined as:
ci, k Rak(1) where
Rc RAN, i, k is the estimation of the resource consumption of user i requesting a service in cell k in the RAN if admitted, and Ra RAN, k is the available resource in cell k in the same RAN. The resource consumption estimation depends on the type of the requested service as well on the radio conditions of the user and the cell load. The available resource is a cell related information and allows the actual traffic load of the cell to be taken into account, as the higher the traffic load, the lower is the available resource at the Base Station, and consequently the higher is the cost.
In the GPRS network, the shared resource R is represented by physical channels, which is the pair (Fc, Ts), where
Fc is the frequency carrier and
Ts is the timeslot number.
The resource consumption of each service supported by the GPRS network is the number of timeslots required to provide the service bit rate, while the available resource at the Base Station is represented by the number of timeslots not yet occupied. Note that the cost definition in (1) covers both channel limited and interference limited cases for GPRS, but the case study presented in this paper covers only the channel limited case. In the UMTS network, the shared resource is the downlink power at the Base Station, therefore the resource consumption is represented by the amount of power required by each service to be satisfactorily provided, and depends on the C/I target associated and the actual traffic load, while the available resource is represented by the amount of power not yet used. It can be noted that differently from the GPRS network the resource consumption is not known a priori; therefore it needs to be estimated by the network in order to calculate the cost. For the I th user in the k th UMTS cell the power consumption estimation is Pest
i, k = maxi(Pj, k)(2)
Where
Pj, k are the downlink powers of MTj which are being provided the same service of MTi in the kth cell. In order to get more power estimates the following procedure can be applied: every downlink power in the cell is normalized with the respective C/I target,
According to the service:
P norm j, k = Pj, k C/I target j (3)
Obtaining for each MT the downlink power needed per C/I unit. Hence the power consumption estimation is performed in this way:
P esti, k= maxj (pnorm, k) ¢ C/I target i (4)
Now all MTs are considered and therefore the estimation is more accurate. This is obviously a conservative procedure as in most cases, when the RAN is loaded, the actual power consumption will be smaller than the estimated one. It can also be noted that the traffic load of the UMTS network is taken into account in both the terms of the fraction: when the load raises the available power decreases and the power consumption estimation increases.
The first TC algorithm is implemented by simply comparing the costs computed by the RANs for each cell and then assigning the new MT to the RAN with the lowest cost. This will be referred to as the basic RAN selection strategy. As the cost definition in UMTS is based on the power consumption, the cost itself depends strongly on the users radio conditions. In fact, MTs at cell borders require more downlink power to get satisfied, and this decreases the cell capacity and raises the cost. Instead, in the GPRS network the quality of service of the generic MT is less sensitive to interference conditions because of frequency reuse. Therefore, if there are enough available channels a user is assumed to be satisfied either at the cell border or near the Base Station, with the same amount of resource expenditure.
The second strategy investigated in this paper tries to move UMTS MTs suffering from bad radio conditions to the GPRS network. This is done once the MT has been admitted in the UTRAN: the actual power consumption is measured and compared to the estimated power used to provide the cost on which the selection was made. If the actual power is greater than the estimated one, it means that this MT is more resource consuming than any MT already served and therefore it is commanded to reselect to GPRS. This approach is expected to be applicable in low mobility environments. This variant of the algorithm is denoted as â„¢TC with reselectionâ„¢. These two schemes have been implemented into two different versions: RAN-based cost updating and cell based cost updating, described above. The RAN-based schemes use information from the whole network to determine the cost to be broadcasted, i.e. the resource consumption estimation is made by considering all MTs in the RAN, and the resource availability is determined averaging the information from all cells.
ADVANTAGES
1. The spacecraft can accommodate any signaling waveform as there is no on-board processing. On-board digital channelization and beam forming consume significant amounts of power which could be used to gain page link margin instead.
2. This strategy allows evolution of signal processing algorithms (such as interference cancellation) via implementations in the ground station. However, ground based beam forming typically requires larger feeder page link bandwidth allocations.
3. Regional based frequency assignment is preferable to satellite based frequency assignment with regard to mobility management. This simplifies the development of the satellite switching center as it is more akin to cellular. In fact in this approach the satellite beams are considered to be the highest layer of the cellular hierarchy and select the one with the lowest cost.
CONCLUSION
Thus by providing a two stage algorithm, the problems regarding traffic can be solved easily. It is clear that providing a service by means of GPRS, which has a limited number of channels, or by means of UMTS, which has a limited Base Station power, implies differences in spectrum efficiency. This simplifies the development of the satellite switching center as it is more akin to cellular. In fact in this approach the satellite beams are considered to be the highest layer of the cellular hierarchy and select the one with the lowest cost.
Then a reselection mechanism has been introduced to improve performance, moving Mobile Terminals which suffer from bad radio conditions, due to high interference level, from UMTS to GPRS. These two schemes have been implemented either calculating the cost on a RAN basis, collecting information from all cells, and on a cell basis. The cell based scheme with reselection was shown to be best one, with a gain of more than 15% against the reference case of RANs run independently, with 10%-15% of Mobile Terminals switched from UMTS to GPRS.
REFERENCES
1. R. Berezdivin, R. Breinig, R. Topp, Next Generation Wireless Communications Concepts and Technologies, IEEE Communication Magazine, March 2002.
2. IST OverDRIVE Project Website http://ist-overdrive
3. .P. Leaves, F. Malavasi, L. Vignali, M. Breveglieri, Description of Traffic Distribution Process Suitable for DRIVE, IST-1999-12515/DRiVE/WP1/D12, March 2002