Coverage & Capacity in Hybrid Wideband Ad-hoc/cellular Access System Results & Scenarios by Pietro Lungaro
Agenda Problem Statement System Assumptions Results Threats
Problem Statement Design a possible networking scenario, and carry out a realistic study about the eventual feasibility, in terms of Coverage & Capacity, of a cellular Ad-hoc multihop network.
Cellular traffic model: open system with Traffic Divergence at the sinks up dn
System Assumptions (Traffic) Even Traffic Distribution among the nodes Poisson Distributed Arrivals of Packets in the transmission buffers ( /N) up Answering Poisson Distributed Traffic from the Basestation towards the terminal nodes ( ) Single service assumption (Implemented also a possible extention to multiservice) B.I.B.O. Stability criterion for the determination of the Maximal Network throughput [Packet/Slot]
Uplink & Downlink traffic specs up is the total (from all the nodes) traffic offered to the base station. In the Uplink side It ll be assumed for i up every node i that: up In the Downlink side It ll be considered an answering traffic equal to: i dn N dn N
B.i.b.o. Stability as a traffic evaluator (1) General description: System Reduction for STDMA
B.i.b.o. Stability as a traffic evaluator (2) Expected Queue Length: L q G' 1 2 2 1 Expected Delay: _ D Fl 2 1 1 where i T f Maximal Node Throughput: i FL 1
System Assumptions (Propagation Model) Standard Model with 4 2 Ptra Ant Ant Sha P rec Gtra Grec Gtra rec L d 4 Shadowing variance set according to the nodes which constitutes the link: Bs-Bs=6 db Bs-Mob=8 db Mob-Mob=12 db d C 10 log d L 10 Frequency carrier 2.4 Ghz, with B=8 Mhz
System Assumptions (Radio Layers) Design & implementation of the Radio Access Layer Solution: Scheduling Routing Reorganizing STDMA (**) MHA Reor-Alg(*) (*) MILLCOMM 85 & Robertazzi (Converted to STDMA) (**) Grönkvist & Somarriba
RESULTS: Throughput Double focus: 1) Users performances (Dimensioning of the Network to improve users rate) 2) Operator Performances (Dimensioning of the Network to reduce the AP density)
RESULTS: Throughput Users Study: Constant Number of active terminals and variable cell radius (Variable Users/Km density) Operator s study: Constant User density and variable number of AP over the same area
User study:example Network for Reuse 1
User study:example Network for Reuse 3
User study:outage Comparison with Service Threshold 5%
Discussion on Throughput The definition given is in [Packet/Slot], or assuming a packet to be enterely transmitted in one slot ([Packet/Packet_duration]). From a B.I.B.O. Stable Queuing Network model is possible to find a bounduary for the network: With this definition of throughput [Packet/slot], assuming a packet duration coincident with the slot duration, it is possible to estimate the Delay [Slots]. Net N / Eff F L Having a figure for the maximal service delay [sec], and knowing coding and Bandwidth it is possible to compute the packet size [bytes] to fulfill the service requirements.
User study:complete Throughput (Packet/Tslot) Comparison (K=1 and K=3) (Each Reuse 3 Packet is assumed to carry a third of Reuse 1 Packet information)
User study:complete Throughput [Kbit/Sec] Comparison (K=1 and K=3)
Operator study: Modeling Service area: 10 km^2 Number of Basestations: 9-16-19-25 Corresponding to 0.9-1.6-1.9-2.5 BS/km^2 Active user density: 15-19-22-26-30-34-38 Active_User/km^2 (or Erlang/km^2)
Operator study: Throughput per site (Bit/Sec/Hz/Site) analysys in function of both Bs-Density & Users-Density
This means: Given that: - The benefits of this architecture are statistically related to the user density. - There can be important Fluctuations in the user density. How can we do a definitive cell planning
Proposed solution for User Density Fluctuations Wireless Router
Routers Scenario: Reuse 1 & variable number of routers per cell 9 Routers 6 Routers 3 Routers
Outage Comparison with Threshold Operator Medium cost Operator Lower cost Operator High cost 5% Network Planning Hysteresis
Outage Comparison with Threshold Medium cost=0.6km Lower cost=0.8km High cost=0.4km Threshold fixed at 5% Need for zoom!!
Routers benefits for Users and Operators: Specific Zoom in the 2-8 Active users Area * = No Routers + = 3 Routers x = 6 Routers o = 9 Routers
Conclusions of the scenario Part of the Infrastructure is now, by definition, an uncontrollable variable, so: These elements could work as traffic stabilizators, correcting in a cheap way errors in the planning or needs to replan (architectural changes). They ll make the performances almost independent from that statistical uncontrollability that can t ensure QoS.
Reorganizing Layer The assumption is that every node in the network has an unique ID number, and related to this there s a reserved initial broadcasting slot.
Reorganizing It is based on a successive inclusion of Subnetworks, one for every level of the network. The Bs is level 0 or Starter. It is based on a priority activation of the links. The metric used is: Layer ( S ) Layer ( ) 2 P ( S, D) I ( S, D) D It works with a Broadcast in the downlink and unicast in the uplink
Reorganizing A reorganizing packet it s much smaller than a data packet (cointains only an estimation of the link qualities). For this reason it could be possible for a node to transmit its info and the relayed ones in a single larger packet (dimensioning the packet size to the largest possible one).
Total Number of Slot Necessary to Reorganize (Variable & Fixed Parts) Comparison
Effective Maximal Number of signaling packets forwarded by an Active Node
Threats (1) The dipendence of the performances on the user density it is Fundamental in a Multi-Hop network. In an environment with high user density fluctuation the cellular planning becames difficult.
Threats (2) The amount of signaling traffic necessary to build and maintain active an Ad-Hoc Network with high QoS needs can conditionate in a negative way the performances; especially when the average hop number increases.
Conclusions With an appropriate cell-planning a Multi-Hop Ad-Hoc solution can ensure interesting data rate to users & a reduction of the base station density to the operators. The prerequisite for all this is an enhanced intelligence in the terminals.