CRS

Congestion in Next Generation Cellular Networks by using Static Channel Relaying Strategy: Analytical Approach Releasing The-Anh Ngo, Sylvie Perreau, and Arek Dadej The Institute for Telecommunications Research, University of South Australia The University of South Australia The.Ngogpostgrads.unisa.edu.au; { Sylvie.Perreau, Arek.Dadej } (unisa.edu.au of channels per cell will be set The main factors used to decide the fixed number of channels for a cell are traffic intensity, SIR, type of services and the QoS requirements Because of its simplicity, FCA is the oldest typical method that can be applied in any cellular network Moreover, it is quite efficient in the cases of heavy traffic during peak hours, achieving high channel utilisation and low call blocking ard dropping probabilities [5] However, FCA has to deal with problems such as user mobility, frequency reuse and interference between cells Dynamic Channel Assignment (DCA) is intended to solve the above problems In DCA, the total number of channels is treated as a pool of common resources available to every user [1]-[3], [5] The channel will be provided in response to a new call request After serving the call, the channel will be freed and retumed to the common pool DCA achieves good load balancing and improves channel utilisation and call blocking probability Furthermore, thanks to the flexibility and adaptivity of channel assignment, the method makes it easy to adapt to users' mobility However, heavy loads and handovers between cells remain serious challenges to DCA Difficult cases of heavy loads and frequent handovers lead to complexities in the operation of DCA Hybrid Channel Assignment (HCA) is a combination of FCA and DCA exploiting their respective advantages [4], [5] In HCA, the network resources are divided into two groups, used for FCA and DCA respectively In the FCA group, channels are assigned to a cell to facilitate serving a fixed number of calls The DCA group is used as the common pool of channels to be distributed on as needed basis This pool will provide channels for new calls in overloaded cells Comparing with DCA and FCA, HCA is likely to be the most effective scheme, achieving good load balancing and satisfying QoS requirements However, the allocation of channels between FCA and DCA groups is a variable dependent on the network and traffic Choosing the right values for particular network is the HCA's challenge Channel relaying is presented as a basic strategy for integrated cellular and ad-hoc relaying (iCAR) systems in [7][91 iCAR system is described as a new architecture for Next Generation Cellular Networks (NGCN), where ad hoc relay Abstact- There have been a number of solutions proposed to deal with congestion resulting from limited number of channels available in cellular networks Channel assignment and channel relaying are two main strategies used to best utilise the available network capacity Static Channel Relaying Strategy (SCRS) is a scheme used to reduce congestion and call blocking probability in hot (overloaded) cells It not only helps in relieving congestion to attain required GoS levels, but also in utilising free channels to achieve load balancing in Next Generation Cellular Networks (NGCN) SCRS can also be used, as proposed in this paper, to localise and minimise hot regions in NGCN to prevent propagation of congestion throughout the network The analytical results on the relaying scheme proposed in this paper show that it is capable of controlling call blocking probability even in the cases of heavy traffic loads in the hot area INTRODUCTION Cellular telephone systems have grown rapidly in the last few decades in both technologies and applications From the first generation (IG) analog voice-only systems, through second generation (2G) digital networks for both voice and data services, cellular networks have evolved to 3G and continue evolving towards 4G, with great improvements in transmission speed, available bandwidth and Quality of Service (QoS) characteristics However, the greatest challenge in these generations of cellular systems is the limited network resources The numbers of subscribers keep increasing every day, while the frequency resource remains limited Solutions to this problem have long been an interesting target for researchers Two main strategies are offered for dealing with the allocation of network resources to cells: channel assignment (CA) schemes [I]-[6] and channel relaying (CR) [7]-[9] The purpose of these strategies is to increase usable network capacity by balancing load across cells and reducing congestion in overloaded cells Fixed, dynamic and hybrid channel assignment schemes (FCA, DCA, and HCA) are the main channel assignment schemes used [5] Fixed Channel Assignment (FCA) refers to the strategy of dividing network resource into nominal channels, which fixed allocation to each cell Depending on the particular network design and traffic forecasts, the number 1-4244-0000-7/05/$20.00 ©2005 IEEE 87 Authorized licensed use limited to: University of South Australia Downloaded on September 3, 2009 at 06:19 from IEEE Xplore Restrictions apply - Degree of a cell d,: This is a value used to indicate the load on a cell via the number of available channels It is calculated as: number- of- available- channels, where Nc is the = dc= N stations (ARSs) are used to relay available channels between cells In iCAR, free channels in cold cells will be "borrowed" by adjacent hot cells via ARS; the relaying routes via ARS are treated as flows of traffic from hot to cold cells In [7], the authors outlined the basic principles of relaying, and the location and numbers of ARSs in the iCAR system The ability of this system to reduce congestion and balance traffic loads among cells is analysed in [8], [9] The analysis shows that significant improvements in load balancing and call blocking probability can be achieved in iCAR system In this paper, we propose a new way to locate ARSs in the NGCN, to maximise the effect (e.g channel utilisation improvement) of relaying Instead of being located at the borders between two cells as in [7], in our proposal the ARSs will be located in the vertices between three cells Thanks to this change of ARSs location, better service will be provided to users, as one ARS can provide access to two neighbouring cells, and there are two ARSs between any pair of cells This will be helpful in localising and minimising the size of hot areas in the NGCN In the paper, we also analyse the load balancing properties of the proposed scheme In summary, we offer a Static Channel Relaying Strategy (SCRS) to release congestion and improve call blocking probability in NGCN The analytical results obtained show that with the use of the proposed scheme, the Grade of Service (GoS) can be kept at values equal to or smaller than 2% for traffic densities in the hot region significantly greater than those allowed by other schemes known from the literature The rest of this paper is organised as follows Section discusses the general concepts and performance of channel relaying strategies in NGCN The application of our proposed SCRS to NGCN is presented in section Analytical results will be given and discussed in section Section will provide summary ofthe paper and conclusions ~~Nc total number of channels in a cell [6] - Threshold of a cell t,: This value is helpful in classifying the cell as hot or cold The typical values of tcare 0.2 and 0.25 [6] The cell threshold is the minimum number of channels required to guarantee that GoS offered by the cell is acceptable, relative to the total number of channels in the cell i.e number- of- guarantee- channels tc - Hot and cold cell: The cell is called hot when d_< t, otherwise it is cold We analyse the performance of channel relaying strategy with the model of three-cell network in Fig.1 as follows: Cellular band SM band Ii GENERAL CONCEPTS AND PERFORMANCE OF CHANNEL RELAYING In practically deployed networks, the traffic load on different cells is not balanced Users will concentrate in certain areas of the network at particular times, e.g around the work and business places during office hours on working days, or around public entertainment areas during weekend The areas where users concentrate may be congested if only a limited number of channels are provided by the corresponding base stations In contrast, some channels in the neighboring cells may be free Consequently, the network resources may be used inefficiently (over-provided in some places, and under-provided in others) Similar to [7]-[9], the term relaying refers here to the method of relaying available channels to the congested areas via ARSs, in order to reduce the congestion and improve call blocking probability in the hot cells Relaying also helps to maximise the overall network capacity by using channels more effectively Before going into detailed study of relaying strategy, we give the following definitions: Figure 1: Channel relaying strategy in NGCN We assume in this model that A is the neighbour cold cell of two hot cells B and C According to the definitions above, we have: * d; > t, in cell A; this means that there are some available channels in this cell, and * d, < t in cell B and C, meaning that B and C are channel hungry cells New calls in these cells may be blocked 88 Authorized licensed use limited to: University of South Australia Downloaded on September 3, 2009 at 06:19 from IEEE Xplore Restrictions apply cell B in order to free up channels for Ax (In Fig 1, relaying routes are set directly from BTS A to B2 via RS2) After relaying, the channels released by B2 can be used for Ax The cells B and C are hot cells, but the traffic can still be relayed between them Consequently, handover relaying is a new (and likely better) mechanism as compared with the dynamic load balancing algorithms in the iCAR system described in [9] In [9], the traffic is only relayed from hot to cold cell while our proposal is applicable to all cases of relying, from hot to cold (static relaying), from cold to hot and between hot cells (handover relaying) In this paper, only the Static Channel Relaying Strategy (SCRS) is studied in detail as a solution that reduces congestion in NGCN The handover relaying strategy will be considered in our future work Because of dC > tc, the free channels in cell A can be exported to cells B and C, which are hot and in need of imported channels to reduce their call blocking probability The channels are relayed by relay stations (RSs) located between the cells RS is a wireless station which operates in the 2.4 GHz ISM band, with the number of channels available for its operation NR The concept of RS is detailed in [7] Suppose that the number of free channels that can be exported from cell A is Ne, and the number of channels which cells B and C have to import to cool down is NJC For successful congestion relief NC=2 NiC has to be assumed We propose that the RSs should be located in the vertices of hexagonal cells The service radius of RS is r = R/2, where R is the radius of a cell This location of RS not only helps the base stations (BTSs) in providing better service to MSs (Mobile Subscribers) in the shadow areas (refer to Fig.1) but also in utilizing the channels for handover relaying (refer to analysis later in this paper) Two main relaying strategies can be used in our proposed scheme as follows Static relaying: The process which provides a channel for the user moving inside the hot cell is called static relaying In Fig.1, suppose that user B I in cell B and user C2 in cell C generate new calls, and B is in the coverage of RS2 while C2 is out of coverage of both RSI and RS3 In the case of B1, the relaying route is set via connections from B I to RS2 and from RS2 to BTS A This type of relying is called direct relaying In order to serve the user C2, RS3 (and also RS1 if necessary) we have to scan to find the user Cl in cell C who is engaged in an on-going call and located within the coverage of one of the two RSs The process of obtaining free channel for C2 is as follows The relaying route will be set between C1 and BTS A via RS3 similar to that between user BI and BTS A above Now the channel released by Cl can be used by C2 This is called indirect relaying The interface between RS and mobile users operates in ISM band, and the interface between RS and BTS uses cellular band to avoid interference [7] Handover relaying: This strategy refers to the relaying process whereby channel is relayed for an on-going call handed over between cells In Fig.I, users Ax in cell A and C3 in cell C are engaged in on-going calls while they move to cell B However, B is a hot cell so there are no free channels in this cell available to serve these calls, hence they would normally be dropped The handover relaying will help to reduce the call dropping probability The handover relaying process can be explained as follows In the case of MS C3, because of its destination in the coverage area of RS4, the relaying route is set to be BTS C -* RS4 -+ MS C3 (Fig.1) In this case, the channel used by MS C3 in BTS C is relayed to the new destination in cell B via RS4 The case of MS Ax is different Because Ax moves out of coverage of the RSs, direct static relaying is carried out for the on-going call B2 in III RELEASING CONGESTION IN THE NEXT GENERATION CELLULAR NETWORKS BY USING STATIC CHANNEL RELAYING STRATEGY Identifying the hot region In this paper, the model of hot region is a hexagonal ring, as follows [6], [9] Hot region is the area where there is a concentration of a fixed number of hot cells within the ring (Fig.2) Hot cells are identified (classified) according to definitions given in Section II In order to localise and minimise the size of hot region, we place a condition on the hot region that there is at least one edge shared between contiguous hot cells in the region Figure 2: The hot region with n = and h, = 0.75 The size and location of hot ring is determined via its diameter d and its center cell In general, d is the largest cell distance between any two cells in the hot area The cell in the middle of the line connecting two most distant hot cells is called the center cell (or ring 0) Ring k is the ring where the 89 Authorized licensed use limited to: University of South Australia Downloaded on September 3, 2009 at 06:19 from IEEE Xplore Restrictions apply distance from the center cell to any cell of that ring is equal k Because of the ring being based on hexagonal cells, the number of cells in the kth ring is equal 6.k Moreover, "ring k" where k > contains corner cells and (6k - 6) non-corner cells (Fig.2) The concepts of corner cells and non-corner cells are detailed in [6], [9] "Ring n", where n = d2, is the largest ring in the hot region Ring (n+1), which covers "ring n", is called "first cold ring" (Fig.2) In order to avoid the spread of congestion and minimise the geographical scope of the operation, the relaying process will be limited to the area covered by the "first cold 3n [Nec (I- d)- Nic dhl+ 3n[4NR + Nec (I d> Nic dhl ±6NR + [Nec (1- dh) Nic dh] O - Let TNC +NK h (NC+N.Tk k=o n IV THE ANALYTICAL RESULTS AND DISCUSSION The number ofrelay stations: The number of relay stations in SCRS is calculated as follows: If the hot region is only I center ring (ring 0), the free channels in cells of the first cold ring will be relayed to ring via six RSs In this case, the number of relay stations N., is (1) NRS = If the hot region consists of n rings including ring (suppose that ring is also hot), NRs is calculated as (see (2) (1)): n-I NRS Nc is the number of channels used to cool down the hot cell, and Nec is the number of channels exported from cold cell Hence, the total number of channels needed to provide for hot cells in n rings is: (3) Nneeded-charmels = dh I[ + 3n(n + 1)] NiC And the total number of free channels in cold cells in a hot region is: = ( dh) [1 + 3n(n + 1)] Ne (4) Ne From (1), (3) and (4), the equation for SCRS is as follows: Recall that le where Th is the traffic load in the hot cell As a result, the congestion and call blocking probability in the hot area are reduced The total number of cells in the hot region is calculated by: N,,,, = 1+ k-1E 6k=1+ 3n(n+ 1) = (5) After relaying, the number of channels available in hot cells is (Nc + N,c), so the call blocking in the cells is: the- number- of- hot- cells, is called degree of hot the- total- cells region This parameter will be used to analyse how the SCRS releases congestion in the NGCN as follows: According to the explanation at the end of Section 111.1 above, the maximum number of channels that can be relayed directly from ring (n+1) to the hot area is: NReayed Max 6* 3NR + (6n-6) -2N we (5) The maximum number of rings in the hot region (the largest size of hot ring) is given by the equation (5) above Hence, nm,x < Max{n,, n2,}, where n, and n2 are the roots of = = (i-dh)- Nc dIhl; have: a)+ 6NR + a 3n2a + 3n(4NR+ ring" In Fig.2, each corner cell and each non-comer cell in ring (n+1) can export directly to one or two cells in "ring n" The corner cells in the inner ring can import channels from cells in the outer ring, and the non-corner cells can import channels from outer cells This assumption will be used in the following section Releasing congestion in NGCN by using Static Channel Relaying Stra tegy (SCRS) The main purpose of SCRS is to provide the highest possible GoS for users in hot cells It also balances the load across cells by effectively utilising free channels SCRS can also be used to minimise the size of hot area and prevent the propagation of congestion in the network The details of SCRS are as follows The parameter dh , d [Nec = = 6+ 1- 3[6.3+(6k-6).2]= 6+,Z(12k+ 6)= k-1 k-1 + 6(n - I)n + 6(n -1) = 2[1 + 3n(n + 1)] - (6n + 2) where [1+3n(n+1)] is the total number of cells in the region = Compared with the results in [71-[9], SCRS not only minimises the size of hot region, but also optimises the number of RSs The relationship between N N NR Nxc traffic load and the size ofhot region Recall that the maximum number of channels relayed from cold cells in the "first cold ring" to hot area is equal to NR, , NReJayed-Ma~x + fN#e.chcannels :-~Nnededd-channels 90 Authorized licensed use limited to: University of South Australia Downloaded on September 3, 2009 at 06:19 from IEEE Xplore Restrictions apply so the number of available channels after relaying is equal to Nc - 2NR (7) Similarly, the number of available channels after relaying in cold cells in hot region is equal to Nc - N,, (8) Suppose that the value of threshold t and the total number of channels per cell Nc are equal to 0.2 and 40 respectively; the cell will be hot when the number of available channels is smaller than or equal to 8, or the number of served channels is equal to or larger than 32 Meanwhile, the traffic load can reach 23.73 Erlangs with the threshold GoS of 2% In order to maintain the GoS at 2%, the number of required channels will be larger than 32 when the traffic load is larger than 23.73 Erlangs (Erlang B formula) The maximum traffic load without relaying is 31 Erlangs (when all of 40 channels are busy) We define that T, Tc = 0.8Th and T = 0.8T,h are the traffic loads on hot cells, cold cells in hot region, and "first cold ring" cells respectively These definitions and equations (7) and (8) can be used to build the relationships between traffic intensity and the size of hot region in the NR.,NJ, case of maintaining GoS at 2% as follows (Fig.3) Th=38.39, Nec =o, NR=3 and Nic=8 dh 0.5 0.75 1- nMax Figure 4: The size of heavy traffic network The maximum number of hot rings, equal to (ring 0, ring 1, ring 2), is given when nMaX = and dh = 0.5 In this case, there are nine hot cells in the nineteen cells heavy traffic region with the corresponding parameters as mentioned above In contrast, when Nic= (1,2,3), Nec= (4,5) and dh= 0.5 or N-c=l, N00 =5 and dh=0.75 respectively (Fig.3), our SCRS proposed scheme can be used to satisfy any size of hot regions (nMaX= 00) Compare with the results in [6] where GoS = 2% when nM,,= 50 and dh= 0.5 We conclude that SCRS proposed here is much better It also can be seen in Fig.3 that the traffic load in hot cell, cold cells in hot region, and first cold ring cells, increase together Consequently, in order to keep GoS at 2% when the traffic load grows, the number of imported channels Nic has to be greater, and the number of exported channels Nec smaller The maximum size of hot region also has to be limited to a smaller value Equation (5) derived for the SCRS scheme proposed in this paper describes the relationship between SCRS's resources (NR, N Nec) and the size of hot region (n) In order to examine the effect of SCRS to maintain GoS at 2%, we study the network model with Nc= 40, Th=36 Erlangs, Nec= I and NR=4 According to (5), after choosing the values of NR and Nec, the relationship between Ni, and n with different values of dh is determined in Fig.5 In general, the size of hot region is limited However, when Ni,=I, dh=0.5, SCRS also be used for relieving congestion in any hot area It can be seen in Fig.5 that the area where SCRS can be applied is smaller when the cells are hotter and the number of required channels greater Recall that when Nc= 40, Th=36.53 Erlangs, Ni,= 6, Ne,= and NR= 4, n will reach 5, and with dh= 0.5, 0.75 and respectively to keep GoS at 2% (Fig.3) However, SCRS can be used in larger area with the same network model as above with different values of GoS, as shown in Fig.6 Max nmax nMax nMax Max of with with with Tc= Th 0.8Th of N, 0.8Tf NR dh=0.5 dh is also used as a condition to identify the cold cells in hot region These conditions (O.< nMax < Max{n1, n2 } and N_C > ) can be used to control the values of Th, NR, and N1, The largest values of Th7, NR and N,C equal to 38.39 Erlangs, and respectively, are chosen to satisfy the condition above Because of Nec= when Th= 38.39 Erlangs, all cells are overloaded or at the threshold of overload These chosen values are an example of heavy traffic Note that the degree of hot region d, is the parameter that is used to indicate the number of hot cells The size of hot region is calculated via different values of db and the network's heavy traffic parameters chosen above as follows: 91 Authorized licensed use limited to: University of South Australia Downloaded on September 3, 2009 at 06:19 from IEEE Xplore Restrictions apply I 35 [1] 30 c [2] _2 e 25 e [3] = 20 p [4] 15 [5] 0 E 10 Z [6] [7] Number of imported channels [8] Figure 5: Imported channels versus number of hot rings [9] REFERENCES A.Pattavina, S.Quadri and V.Trecordi, "Reuse partitioning in cellular networks with dynamic channel allocation", Wireless Networks, Vol.5 Issue 4, pp.299-309, July 1999 S-M.Senouci and G.Pujoile, "Dynamic Channel Assignment in Cellular Networks: A Reinforcement Leaming Solution", IEEE ICT2003, Vol 1, pp.302-309, March 2003 A.Hac and C.Mo, "Dynamic Channel Assigrunent in Wireless Communication networks", International Journal of Network Management, Vol.9 Issue 1, pp.44-66, March 1999 K.A.Agha, "Hybrid Channel Assignment Using Dynamic Resource Sharing", IEEE VTC 1999-Fall, Vol.5, pp.3029-3033, Sept 1999 I.Katzela and M.Naghshineh, "Channel Assignment Schemes for Cellular Mobile Telecommunication Systems: A Comprehensive Survey", IEEE Personal.Comm, pp 10-31, June 1996 S.K.Das, S.K.Sen and R.Jayaram, "A novel load balancing scheme for the tele-traffic hot spot problem in cellular networks", Wireless Networks, Vol.4 Issue 4, pp.325-340, July 1998 C.Qiao and H.Wu, "iCAR: an Integrated Cellular and Ad-hoc Relay System", Proc.9? Int.ConflEEE Computer Comm and Networks, Vol.2, pp.154-161, Oct.2000 H.Wu, C.Qiao, S.De and O.Tonguz, "Integrated Cellular and Ad Hoc Relaying Systems: iCAR", IEEE journal on Selected Areas in Comnications, Vol.19 Issue 10, pp.2105-2115, Oct.2001 E.Yanmaz, O.K.Tonguz, S.Mishra, H.Wu and C.Qiao, "Efficient Dynamic Load Balancing Algorithms using iCAR Systems: A Generpl Framework", in Proc.IEEE VTC 2002-FaU, Vol.1, pp.586-590, Sept.2002 Figure 6: Call blocking probability vs size of hot region V CONCLUSION In this paper, Static Channel Relaying Strategy is proposed as an efficient strategy to deal with the limitations on the number of channels in Next Generation Cellular Network The strategy not only reduces the congestion and call blocking probability, but also effectively utilises free channels in the network via relay stations located in the vertices of hexagonal cells SCRS can also be used to localise and minimise the size of a hot area The analytical results demonstrate that good GoS levels can be achieved even in heavy traffic and for large hot areas 92 Authorized licensed use limited to: University of South Australia Downloaded on September 3, 2009 at 06:19 from IEEE Xplore Restrictions apply  ... Stra tegy (SCRS) The main purpose of SCRS is to provide the highest possible GoS for users in hot cells It also balances the load across cells by effectively utilising free channels SCRS can also... derived for the SCRS scheme proposed in this paper describes the relationship between SCRS's resources (NR, N Nec) and the size of hot region (n) In order to examine the effect of SCRS to maintain... (Fig.3), our SCRS proposed scheme can be used to satisfy any size of hot regions (nMaX= 00) Compare with the results in [6] where GoS = 2% when nM,,= 50 and dh= 0.5 We conclude that SCRS proposed