"The power usage of OFDM-enabled portable devices still leaves something to be desired, too." Power consumption of delivering very high bandwidths is an issue, particularly for the up-link where antenna gain is limited. MIMO and AAS or MIMO-AAS combinations will be used to help increase SNR, reduce interference and allow frequency reuse. LTE is also addressing this problem with the use of SC-OFDM for the uplink while also using OFDMA for the down link.
The goal for 4G as set out by ITU is to deliver 100 Mbps for mobile use and up to 1 Gbps for fixed-nomadic use. The only way in the world to deliver that is to have localized networks connected very granular into high bandwidth fiber optic or very high bandwidth wireless back-haul (cable can be used but if you are going to go to the trouble of running new lines, fiber optic kicks cable's butt). So, this sets up a paradox: to build a wire area system that delivers mobility you have to design user devices that will use low power. But if requirement trends continue, such as person-to-person viral video, self-recorded playback video (Tivo and new Apple TV for example), p-t-p file sharing and myriad of applications we haven't thought of yet, then networks will have to be deployed with very high density. The higher the density, the more advantage can be taken using MIMO to increase SNR and reduce power requirements.
I also think that we will see more evolution to 'multi-spectrum' devices. Qualcomm is still the best strategic thinker in the world imo: they took advantage of advances in chip, antenna and other designs to fashion MediaFlo OFDM plus 3G CDMA into a multi-service, multi-spectrum enabling chip set, system and service. That is way ahead of most folks thinking about WiMAX or the limitations of WiFi.
The discussions about WiFi being able to do much of what has been considered for WiMAX are transitory: WiFi has very limited spectrum which is also critically constrained by being contention based. It can and will continue to evolve to do more but the scope is inevitably limited. WiMAX has a much broader scope and it can be used in localized as well as multiple spectrums and types of services. The pitfall is that this takes vision and dedicated development, some of which appears to be scarce among the many companies involved.
'This is a market that will sacrifice QOS for price" Not all customers are alike. Skype usage has greatly tapered off because they can't get around the problems of poor and consistent QoS. Do millions use it? Yes, but many hundreds of millions more each year pay for mobile and more reliable wired and wireless service. Cheap s not necessarily very appealing... even in third world countries. The 'right price' is what sells and what propels development and deployments.
Femtocells are another question: depending on range, we should also look at UWB as an alternative. UWB is limited to around 20 meters. I think it can be useful for PANs but there is overlap with WiFi which can use power saving/scaling techniques to reduce power levels. Many PAN devices can be line-powered, reducing the reliance on battery power where power reduction if most important. WiMAX overlaps moe with WiFi than UWB PANs or 'wireless USB' but, this will be made multi-mode and multi-spectrum.
What enhances the long term outlook for these trends is the developments in semiconductor devices: the use of High K dielectric and new metal interconnect techniques is the first major shift from CMOS for very highly integrated mass market chips over the past 30 years. Intel will begin using these processes in processors starting the middle of this year. It will probably be 2-4 years before we see the impact in wireless SoCs. A major impact will be reduction in power. Immediately a lot of power can be saved in the processors that run laptops and PDAs. But this can be extended to chips that integrate wireless on-chip or in discrete chips as well. There are additional advantages in lower SNR and higher RF sensitivity that is similar to the use exotic processes such as gallium arsenide which is much more costly and more difficult to do SoCs. As far as I know, Intel has not announced specific plans to implement WiMAX or multi-mode SoCs using the new processes but they have demonstrated lab chips with multi-mode capability. I think it is inevitable they will use the new processes once they have gained experience and learn how to use them for RF and other analog functions. But that is a lot my own speculation at this point as you can't necessarily rely on demos or trends. |
