I attended IEEE PIMRC’13 conference held at London last week in order to present a paper about 5 GHz dense Wi-Fi capacity. Overall, the conference itself was very useful to get updated both from industries and academia. From key note speeches (See below the post by Ki Won) and two panel discussions about 5G, there seems at least consensus among vendors and operators on why we need 5G, i.e., massive data volume & connectivity, and ‘true’ ubiquitous services in reality (not in wireless textbooks and articles). I think no ones would disagree the 5G system should support order of magnitude more capacity and connectivity than LTE and its advance.
Who will invest a network for what? Content ownership vs. Network ownership
From investment perspective which is a necessary condition for such high-speed mobile Internet access, as Stephen Pusey, CTO of Vodafone, mentioned in a key note speech, cloud based and video services + personalization will be one strong driver although it is still questionable if it will dominantly happen and replace currently dominant voice+low to moderate basic mobile broadband services. A key uncertainty is on whether or not telecom operators are willing to invest actively for such services provided by Google or Facebook. From content provider perspectives, they earn money from social infrastructure and hope it will be extended more and more without their explicit investment from their pockets (the wide spread of wi-fi access points would be more preferred by content providers, as discussed in Spectrum Sharing panel discussion, see the post below by Jan). However, without new service models by telecom operators, existing mobile broadband markets sooner or later will reach 100% of penetration rate so that they would more passively invest a network in order not to lose their subscribers. One scenario could be that operators directly provide those services although they mostly lack high-quality of contents and brand names compared with huge existing content providers, e.g., YouTube or Facebook. Alternatively, those content providers may tightly cooperate with operators.
New commercial system or New standard?
Let’s assume that someone (either telcos or new future actors) will invest a network (this is one key underlying assumption for 5G system). With the clear drivers of more capacity and billions of connections, I personally believe that new commercial systems are needed at least due to two reasons. First, the maintaining older generation systems could be more expensive 10 to 20 years later than deploying more efficient new systems. This normally happens in most of other infrastructure businesses. Secondly, the advances of hardware and electronics may make some technologies feasible 10 years later which may not be feasible today. From a research point of view, those issues mainly are about numerology, optimizations, and hardware development (computing powers and chip integration, etc) than network architecture issues which are closely related to standardization which 5G will be normally defined from. It may not be super clear to me that the requirements only from scale and number will need new standard. Historically, as far as I understood, there were high-level big jumps between wireless system generations. Let’s look at what were major change in the older generation systems. 1G (AMPS) to 2G (GSM): analog to digital, 2G to 3G (WCDMA,HSPA+): circuit switched to added packet switched, 3G to 4G (LTE(-A)): centralized hierarchical architecture –> more flat and distributed architecture, all IP based. For 5G, the high-level new system concept which makes people easily agree to require new standards may still need to be developed. This could be one role of research community.