Transforming 3G radio Access Architecture Ionut BIBAC & Emmanuel DUJARDIN Agenda Main Triggers for New Access Architecture Toward Flat Architecture: Issue and Limitation The 3M of beyond 3G: Multi-Carrier, Multi-Antenna (MiMo) and Multi-Layer One Word on SDR… Conclusion Main Triggers for Deploying New Access Architecture Access to a larger (and variable) spectrum allocation Higher spectrum efficiency which implies: Reduction latency with a better QoS and user experience Variable channel BW and harmonized FDD/TDD enables greater flexibility to exploit different band allocations. Spectrum reframing where we can take advantage of the flexible channel BW and/or better potential use of TDD spectrum. Optimized for flat architecture (should leave to lower cost network in the long term) Not burdened by need to support legacy terminals and protocols leads to optimized spectrum efficiency and latency performance. Higher capacity per site should lead to lower cost/bit at high traffic levels. Capability to support new service and/or competition with other technologies that requires the lower latency of LTE to achieve good/equivalent customer satisfaction. Towards Flat Architecture flat architecture fewer layers of network elements (collapsed architectures) fewer central bottlenecks more any to any connectivity drivers / expected benefits (to be confirmed) costs: lot of small not redundant units cheaper than few central high capacity, highly reliable network elements (including hosting costs) performance: traffic go through fewer equipments, more direct routes => less latency, jitter, better throughput Examples: LTE/EPC HSPA flat architecture / I-HSPA Direct Tunnel femtocells 3G – LTE/EPC – HSPA Flat PSTN MSC 3G: HLR NB RNC LTE/EPC (3GPP R8): SGSN GGSN MME HLR Serving Gateway eNB PDN Gateway Data Data HSPA Flat Architecture (3GPP R8 option) / I-HSPA: RNC PSTN MSC HLR NB/RNC SGSN GGSN Data Direct Tunnel - Femto Direct Tunnel: (3GPP R7) NB Direct Tunnel + HSPA Flat: PSTN MSC HLR SGSN RNC RNC GGSN Data PSTN MSC HLR SGSN NB/RNC GGSN Femto (not standard yet): HNB= ~NB/RNC FGW Data PSTN MSC HLR SGSN GGSN Data Issue and Limitations of Flat Architecture data only (except femto*): if voice on circuit, feasibility and performance to be checked (for example on I-HSPA): About Femto: most issues are currently handled with a gateway/proxy that hides complexity from CN…but not really flat..though collapsed signalling: all mobility is managed at CN level => either CN correctly designed to handle it (EPC?) or best fitted for slow moving users Security: any to any connectivity assumes IP transport network, could be 3rd party network or even public internet collapsing RNC functions into NB involves that radio ciphering is done in NB direct connection to CN equipments (except femto*) impact on existing equipments (configuration and interface): more network nodes visible (except femto*) interworking and interconnections to legacy architectures need to have a centralized point of interconnection The 3M of beyond 3G: Multi-Carrier OFDM basic principles Carrier (e.g. 5 MHz) is subdivided into many narrower band sub-carriers with lower rates User receives many sub carriers together to achieve higher rates Designed to achieve low distortion on each sub-carrier due to radio reflections and adjacent sub-carriers –5 MHz Bandwidth –FFT –Sub-carriers –Guard Intervals –… –Symbols –Frequency –… –Time The 3M of beyond 3G: Multi-Antenna (Mimo) MIMO = Canal matriciel xi : Signaux émis Yi : Signaux reçus N canaux de' transmission parallèles x1 y1 x2 y2 xN yN Problème du récepteur: Retrouver signaux émis X X = H-1 Y Possible si H est inversible Eléments hij décorrélés Conditions les plus favorables: Milieu très réflectif Plutôt Indoor The 3M of beyond 3G: Multi-Layer One Word on SDR… Software Defined Radio stands for a radio technology agnostic Hardware platform in which some or all Radio and Baseband functionalities are controlled by Software. Early GSM specifications, about filters and frequency blocking, are challenging.Some demand for relaxation of the band. Difficulty to precisely estimate today the necessary processing power for a later use, towards LTE for instance and ultimately any other new usage. Coexistence of technologies in same modules is not easy to manage. Vendors are tied with their current chipset choices. Moving to fully SW defined platform means initially full re-development of firmware. On the other hand they gain full flexibility on future development. Roadmaps shows no OSS evolution with SDR introduction. For instance, changing the technology is done by deletion & recreation of cells, all of the earlier settings and optimisations are lost. SDR cannot (yet) be considered as a dynamic configuration enabler Conclusion Operator benefits of the new air interface Access to larger (and variable) spectrum allocations Higher spectrum efficiency: lower cost per bit Reduced latency: better QoS ans user experience Reasons for migration Higher spectrum efficiencies can also be achieved by HSPA+ with lower migration cost (assuming 5 MHz spectrum allocation) New spectrum allocations or re-farming may motivate migration (currently 20 MHz allocations seem very unlikely but 10 MHz may be possible) E-UTRAN will be deployed together with evolved packet core (EPC) Air Interface evolution will continue IMT advanced seems far away for operators. Concurrent systems are in starting blocks so 3GPP also has to respond.