As the wireless industry plays an ever larger role in the global communications infrastructure, the level of development effort going into the evolution of different wireless technologies is remarkable. On June 12 in Dallas, I chaired a Portable Computer and Communications Association (PCCA) meeting on HSPA (High Speed Packet Access) sponsored by 3G Americas. On June 13, AT&T hosted an HSPA workshop under an NDA arrangement where attendees were able to get hands-on time with forthcoming services and devices. The meeting provided the perfect crystal ball experience to see how wireless networks will be able to deliver speeds almost one hundred times faster than current 3G systems by early next decade. That's remarkable when you consider that current 3G systems are a hundred times faster than the first wireless IP systems such as CDPD (Cellular Digital Packet Data).

HSPA is the name used for UMTS (Universal Mobile Telecommunications System) networks that provide both HSDPA (High Speed Downlink Packet Access) and HSUPA (High Speed Uplink Packet Access). A presentation from 3G Americas included market statistics showing 169 UMTS operators in 71 countries and 117 HSDPA operators in 59 countries. Clearly, HSDPA is seeing a global surge of deployment, and for good reason. With user rates frequently above 1 Mbps and sub-100 msec latency, most networking applications work extremely well. As explained in a presentation from Ericsson, HSDPA was specified in 3GPP (Third Generation Partnership Project) Release 5 specifications. Release 6 includes HSUPA, and you'll start seeing HSUPA this year from UMTS operators such as AT&T. HSUPA will bring uplink speeds in line with downlink speeds, though as with EV-DO Rev A and WiMAX, uplink spectral efficiency is typically only half of downlink spectral efficiency, so peak rates normally will be lower than the downlink rate. Nevertheless, uplink speeds will be impressive. Ericsson showed live operator data for one network with median bit rates of 1.0 Mbps. HSUPA also enables lower latencies, as low as 50 msec when measured from the mobile device to the edge of the network. HSDPA itself will get faster, with devices supporting peak rates of 7.2 Mbps, enabling real-world throughputs of 2 Mbps to 3 Mbps, assuming the operator has the bandwidth in its backhaul to support these rates.

Partly in response to competitive pressure from WiMAX, HSPA innovators have included significant new HSPA capabilities in Release 7 and further refinements in Release 8. Referred to as either HSPA+ or HSPA Evolution, and planned for availability in 2008, peak theoretical rates will increase to 28 Mbps (from the current peak theoretical rate of 14.4 Mbps) and then to 42 Mbps. The key enabling mechanisms are the addition of 64-QAM modulation (HSDPA currently uses 16-QAM for maximum throughputs) and MIMO (multiple input multiple output), initially at 2X2. In the future, 4X4 MIMO could boost rates all the way to 80 Mbps. The goal of HSPA+ is to match OFDMA (orthogonal frequency division multiple access) approaches, assuming operation in 5+5 MHz (5 MHz down, 5 MHz up) of spectrum. WiMAX may still outperform HSPA+ in larger spectrum bands or in looser frequency reuse configurations, but that takes large amounts of free spectrum that most operators simply don't have.

As a presentation from Alcatel Lucent demonstrated, things really start to get interesting with 3GPP LTE (3GPP Long Term Evolution), which 3GPP specifies in Release 8. LTE uses OFDMA like WiMAX. Being a newer technology, however, it incorporates enhancements that WiMAX will only be able to match (or exceed) with the IEEE 802.16m specification--which is now only in the early stages. Intended to be available in 2009, the LTE specification will be completed in early 2008. There is a lot to say about LTE, but I'll just mention that peak theoretical rates hit a blistering 326 Mbps in 20 MHz of spectrum with 4X4 MIMO. Real-world throughputs will obviously be lower, and widespread adoption won't happen until sometime next decade, but that's the roadmap, and it's impressive.

Don't count out GSM and EDGE, which still account for the bulk of global subscribers. RIM presented on Evolved EDGE, now called EDGE II. Expected around 2008, EDGE II will boost speeds to around 1 Mbps by adding new modulation, mobile receive diversity and modems communicating on more than just one radio channel simultaneously--as well as using many more time slots than current EDGE devices. Evolving EDGE makes sense because it allows some operators to defer expensive 3G investments. For 3G operators, Evolving EDGE enables hybrid 3G/EDGE systems, where running on EDGE say in a rural area doesn't mean a huge step down in performance.

In short, 3GPP's three-prong approach of EDGE Evolution, HSPA Evolution and 3GPP Long Term Evolution yields an impressive roadmap of continued performance increases.

Peter Rysavy is the president of Rysavy Research (http://www.rysavy.com/), a consulting firm that specializes in wireless technology assessment and integration.